Cihang Xie

Yutong Bai, Zeyu Wang, Junfei Xiao et al. · berkeley

This paper studies the potential of distilling knowledge from pre-trained models, especially Masked Autoencoders. Our approach is simple: in addition to optimizing the pixel reconstruction loss on masked inputs, we minimize the distance between the intermediate feature map of the teacher model and that of the student model. This design leads to a computationally efficient knowledge distillation framework, given 1) only a small visible subset of patches is used, and 2) the (cumbersome) teacher model only needs to be partially executed, ie, forward propagate inputs through the first few layers, for obtaining intermediate feature maps. Compared to directly distilling fine-tuned models, distilling pre-trained models substantially improves downstream performance. For example, by distilling the knowledge from an MAE pre-trained ViT-L into a ViT-B, our method achieves 84.0% ImageNet top-1 accuracy, outperforming the baseline of directly distilling a fine-tuned ViT-L by 1.2%. More intriguingly, our method can robustly distill knowledge from teacher models even with extremely high masking ratios: e.g., with 95% masking ratio where merely TEN patches are visible during distillation, our ViT-B competitively attains a top-1 ImageNet accuracy of 83.6%; surprisingly, it can still secure 82.4% top-1 ImageNet accuracy by aggressively training with just FOUR visible patches (98% masking ratio). The code and models are publicly available at https://github.com/UCSC-VLAA/DMAE.

Jieru Mei, Yucheng Han, Yutong Bai et al. · berkeley

Adversarial Propagation (AdvProp) is an effective way to improve recognition models, leveraging adversarial examples. Nonetheless, AdvProp suffers from the extremely slow training speed, mainly because: a) extra forward and backward passes are required for generating adversarial examples; b) both original samples and their adversarial counterparts are used for training (i.e., 2$\times$ data). In this paper, we introduce Fast AdvProp, which aggressively revamps AdvProp's costly training components, rendering the method nearly as cheap as the vanilla training. Specifically, our modifications in Fast AdvProp are guided by the hypothesis that disentangled learning with adversarial examples is the key for performance improvements, while other training recipes (e.g., paired clean and adversarial training samples, multi-step adversarial attackers) could be largely simplified. Our empirical results show that, compared to the vanilla training baseline, Fast AdvProp is able to further model performance on a spectrum of visual benchmarks, without incurring extra training cost. Additionally, our ablations find Fast AdvProp scales better if larger models are used, is compatible with existing data augmentation methods (i.e., Mixup and CutMix), and can be easily adapted to other recognition tasks like object detection. The code is available here: https://github.com/meijieru/fast_advprop.

Yiqing Wang, Zihan Li, Jieru Mei et al. · uw

Recent advancements in large-scale Vision Transformers have made significant strides in improving pre-trained models for medical image segmentation. However, these methods face a notable challenge in acquiring a substantial amount of pre-training data, particularly within the medical field. To address this limitation, we present Masked Multi-view with Swin Transformers (SwinMM), a novel multi-view pipeline for enabling accurate and data-efficient self-supervised medical image analysis. Our strategy harnesses the potential of multi-view information by incorporating two principal components. In the pre-training phase, we deploy a masked multi-view encoder devised to concurrently train masked multi-view observations through a range of diverse proxy tasks. These tasks span image reconstruction, rotation, contrastive learning, and a novel task that employs a mutual learning paradigm. This new task capitalizes on the consistency between predictions from various perspectives, enabling the extraction of hidden multi-view information from 3D medical data. In the fine-tuning stage, a cross-view decoder is developed to aggregate the multi-view information through a cross-attention block. Compared with the previous state-of-the-art self-supervised learning method Swin UNETR, SwinMM demonstrates a notable advantage on several medical image segmentation tasks. It allows for a smooth integration of multi-view information, significantly boosting both the accuracy and data-efficiency of the model. Code and models are available at https://github.com/UCSC-VLAA/SwinMM/.

Zeyu Wang, Yutong Bai, Yuyin Zhou et al. · berkeley

The recent success of Vision Transformers is shaking the long dominance of Convolutional Neural Networks (CNNs) in image recognition for a decade. Specifically, in terms of robustness on out-of-distribution samples, recent research finds that Transformers are inherently more robust than CNNs, regardless of different training setups. Moreover, it is believed that such superiority of Transformers should largely be credited to their self-attention-like architectures per se. In this paper, we question that belief by closely examining the design of Transformers. Our findings lead to three highly effective architecture designs for boosting robustness, yet simple enough to be implemented in several lines of code, namely a) patchifying input images, b) enlarging kernel size, and c) reducing activation layers and normalization layers. Bringing these components together, we are able to build pure CNN architectures without any attention-like operations that are as robust as, or even more robust than, Transformers. We hope this work can help the community better understand the design of robust neural architectures. The code is publicly available at https://github.com/UCSC-VLAA/RobustCNN.

Haoqin Tu, Chenhang Cui, Zijun Wang et al. · oxford

This work focuses on the potential of Vision LLMs (VLLMs) in visual reasoning. Different from prior studies, we shift our focus from evaluating standard performance to introducing a comprehensive safety evaluation suite, covering both out-of-distribution (OOD) generalization and adversarial robustness. For the OOD evaluation, we present two novel VQA datasets, each with one variant, designed to test model performance under challenging conditions. In exploring adversarial robustness, we propose a straightforward attack strategy for misleading VLLMs to produce visual-unrelated responses. Moreover, we assess the efficacy of two jailbreaking strategies, targeting either the vision or language component of VLLMs. Our evaluation of 21 diverse models, ranging from open-source VLLMs to GPT-4V, yields interesting observations: 1) Current VLLMs struggle with OOD texts but not images, unless the visual information is limited; and 2) These VLLMs can be easily misled by deceiving vision encoders only, and their vision-language training often compromise safety protocols. We release this safety evaluation suite at https://github.com/UCSC-VLAA/vllm-safety-benchmark.

Chen Wei, Karttikeya Mangalam, Po-Yao Huang et al. · meta-ai

There has been a longstanding belief that generation can facilitate a true understanding of visual data. In line with this, we revisit generatively pre-training visual representations in light of recent interest in denoising diffusion models. While directly pre-training with diffusion models does not produce strong representations, we condition diffusion models on masked input and formulate diffusion models as masked autoencoders (DiffMAE). Our approach is capable of (i) serving as a strong initialization for downstream recognition tasks, (ii) conducting high-quality image inpainting, and (iii) being effortlessly extended to video where it produces state-of-the-art classification accuracy. We further perform a comprehensive study on the pros and cons of design choices and build connections between diffusion models and masked autoencoders.

Junyang Wu, Xianhang Li, Chen Wei et al.

This paper presents a simple and effective visual prompting method for adapting pre-trained models to downstream recognition tasks. Our method includes two key designs. First, rather than directly adding together the prompt and the image, we treat the prompt as an extra and independent learnable component. We show that the strategy of reconciling the prompt and the image matters, and find that warping the prompt around a properly shrinked image empirically works the best. Second, we re-introduce two "old tricks" commonly used in building transferable adversarial examples, i.e., input diversity and gradient normalization, into visual prompting. These techniques improve optimization and enable the prompt to generalize better. We provide extensive experimental results to demonstrate the effectiveness of our method. Using a CLIP model, our prompting method sets a new record of 82.8% average accuracy across 12 popular classification datasets, substantially surpassing the prior art by +5.6%. It is worth noting that this prompting performance already outperforms linear probing by +2.1% and can even match fully fine-tuning in certain datasets. In addition, our prompting method shows competitive performance across different data scales and against distribution shifts. The code is publicly available at https://github.com/UCSC-VLAA/EVP.

Sucheng Ren, Huiyu Wang, Zhengqi Gao et al.

Data mixing (e.g., Mixup, Cutmix, ResizeMix) is an essential component for advancing recognition models. In this paper, we focus on studying its effectiveness in the self-supervised setting. By noticing the mixed images that share the same source images are intrinsically related to each other, we hereby propose SDMP, short for $\textbf{S}$imple $\textbf{D}$ata $\textbf{M}$ixing $\textbf{P}$rior, to capture this straightforward yet essential prior, and position such mixed images as additional $\textbf{positive pairs}$ to facilitate self-supervised representation learning. Our experiments verify that the proposed SDMP enables data mixing to help a set of self-supervised learning frameworks (e.g., MoCo) achieve better accuracy and out-of-distribution robustness. More notably, our SDMP is the first method that successfully leverages data mixing to improve (rather than hurt) the performance of Vision Transformers in the self-supervised setting. Code is publicly available at https://github.com/OliverRensu/SDMP

Sizhe Chen, Zhehao Huang, Qinghua Tao et al.

The score-based query attacks (SQAs) pose practical threats to deep neural networks by crafting adversarial perturbations within dozens of queries, only using the model's output scores. Nonetheless, we note that if the loss trend of the outputs is slightly perturbed, SQAs could be easily misled and thereby become much less effective. Following this idea, we propose a novel defense, namely Adversarial Attack on Attackers (AAA), to confound SQAs towards incorrect attack directions by slightly modifying the output logits. In this way, (1) SQAs are prevented regardless of the model's worst-case robustness; (2) the original model predictions are hardly changed, i.e., no degradation on clean accuracy; (3) the calibration of confidence scores can be improved simultaneously. Extensive experiments are provided to verify the above advantages. For example, by setting $\ell_\infty=8/255$ on CIFAR-10, our proposed AAA helps WideResNet-28 secure 80.59% accuracy under Square attack (2500 queries), while the best prior defense (i.e., adversarial training) only attains 67.44%. Since AAA attacks SQA's general greedy strategy, such advantages of AAA over 8 defenses can be consistently observed on 8 CIFAR-10/ImageNet models under 6 SQAs, using different attack targets, bounds, norms, losses, and strategies. Moreover, AAA calibrates better without hurting the accuracy. Our code is available at https://github.com/Sizhe-Chen/AAA.

Xianhang Li, Zeyu Wang, Cihang Xie

The recent work CLIPA presents an inverse scaling law for CLIP training -- whereby the larger the image/text encoders used, the shorter the sequence length of image/text tokens that can be applied in training. This finding enables us to train high-performance CLIP models with significantly reduced computations. Building upon this work, we hereby present CLIPA-v2 with two key contributions. Technically, we find this inverse scaling law is also applicable in the finetuning stage, enabling further reduction in computational needs. Empirically, we explore CLIPA at scale, extending the experiments up to the H/14 model with ~13B image-text pairs seen during training. Our results are exciting -- by only allocating a budget of \$10,000, our CLIP model achieves an impressive zero-shot ImageNet accuracy of 81.1%, surpassing the prior best CLIP model (from OpenCLIP, 80.1%) by 1.0% and meanwhile reducing the computational cost by ~39X. Moreover, with an additional investment of $4,000, we can further elevate the zero-shot ImageNet accuracy to 81.8%. Our code and models are available at https://github.com/UCSC-VLAA/CLIPA.

Xianhang Li, Huiyu Wang, Chen Wei et al.

Image pre-training, the current de-facto paradigm for a wide range of visual tasks, is generally less favored in the field of video recognition. By contrast, a common strategy is to directly train with spatiotemporal convolutional neural networks (CNNs) from scratch. Nonetheless, interestingly, by taking a closer look at these from-scratch learned CNNs, we note there exist certain 3D kernels that exhibit much stronger appearance modeling ability than others, arguably suggesting appearance information is already well disentangled in learning. Inspired by this observation, we hypothesize that the key to effectively leveraging image pre-training lies in the decomposition of learning spatial and temporal features, and revisiting image pre-training as the appearance prior to initializing 3D kernels. In addition, we propose Spatial-Temporal Separable (STS) convolution, which explicitly splits the feature channels into spatial and temporal groups, to further enable a more thorough decomposition of spatiotemporal features for fine-tuning 3D CNNs. Our experiments show that simply replacing 3D convolution with STS notably improves a wide range of 3D CNNs without increasing parameters and computation on both Kinetics-400 and Something-Something V2. Moreover, this new training pipeline consistently achieves better results on video recognition with significant speedup. For instance, we achieve +0.6% top-1 of Slowfast on Kinetics-400 over the strong 256-epoch 128-GPU baseline while fine-tuning for only 50 epochs with 4 GPUs. The code and models are available at https://github.com/UCSC-VLAA/Image-Pretraining-for-Video.

Qijun Han, Haoqin Tu, Zijun Wang et al.

Autonomous GUI agents face two fundamental challenges: early stopping, where agents prematurely declare success without verifiable evidence, and repetitive loops, where agents cycle through the same failing actions without recovery. We present VLAA-GUI, a modular GUI agentic framework built around three integrated components that guide the system on when to Stop, Recover, and Search. First, a mandatory Completeness Verifier enforces UI-observable success criteria and verification at every finish step -- with an agent-level verifier that cross-examines completion claims with decision rules, rejecting those lacking direct visual evidence. Second, a mandatory Loop Breaker provides multi-tier filtering: switching interaction mode after repeated failures, forcing strategy changes after persistent screen-state recurrence, and binding reflection signals to strategy shifts. Third, an on-demand Search Agent searches online for unfamiliar workflows by directly querying a capable LLM with search ability, returning results as plain text. We additionally integrate a Coding Agent for code-intensive actions and a Grounding Agent for precise action grounding, both invoked on demand when required. We evaluate VLAA-GUI across five top-tier backbones, including Opus 4.5, 4.6 and Gemini 3.1 Pro, on two benchmarks with Linux and Windows tasks, achieving top performance on both (77.5% on OSWorld and 61.0% on WindowsAgentArena). Notably, three of the five backbones surpass human performance (72.4%) on OSWorld in a single pass. Ablation studies show that all three proposed components consistently improve a strong backbone, while a weaker backbone benefits more from these tools when the step budget is sufficient. Further analysis also shows that the Loop Breaker nearly halves wasted steps for loop-prone models.

Yipeng Gao, Zeyu Wang, Wei-Shi Zheng et al.

Contrastive learning has emerged as a promising paradigm for 3D open-world understanding, i.e., aligning point cloud representation to image and text embedding space individually. In this paper, we introduce MixCon3D, a simple yet effective method aiming to sculpt holistic 3D representation in contrastive language-image-3D pre-training. In contrast to point cloud only, we develop the 3D object-level representation from complementary perspectives, e.g., multi-view rendered images with the point cloud. Then, MixCon3D performs language-3D contrastive learning, comprehensively depicting real-world 3D objects and bolstering text alignment. Additionally, we pioneer the first thorough investigation of various training recipes for the 3D contrastive learning paradigm, building a solid baseline with improved performance. Extensive experiments conducted on three representative benchmarks reveal that our method significantly improves over the baseline, surpassing the previous state-of-the-art performance on the challenging 1,156-category Objaverse-LVIS dataset by 5.7%. The versatility of MixCon3D is showcased in applications such as text-to-3D retrieval and point cloud captioning, further evidencing its efficacy in diverse scenarios. The code is available at https://github.com/UCSC-VLAA/MixCon3D.

Qingyue Wei, Lequan Yu, Xianhang Li et al.

Medical imaging has witnessed remarkable progress but usually requires a large amount of high-quality annotated data which is time-consuming and costly to obtain. To alleviate this burden, semi-supervised learning has garnered attention as a potential solution. In this paper, we present Meta-Learning for Bootstrapping Medical Image Segmentation (MLB-Seg), a novel method for tackling the challenge of semi-supervised medical image segmentation. Specifically, our approach first involves training a segmentation model on a small set of clean labeled images to generate initial labels for unlabeled data. To further optimize this bootstrapping process, we introduce a per-pixel weight mapping system that dynamically assigns weights to both the initialized labels and the model's own predictions. These weights are determined using a meta-process that prioritizes pixels with loss gradient directions closer to those of clean data, which is based on a small set of precisely annotated images. To facilitate the meta-learning process, we additionally introduce a consistency-based Pseudo Label Enhancement (PLE) scheme that improves the quality of the model's own predictions by ensembling predictions from various augmented versions of the same input. In order to improve the quality of the weight maps obtained through multiple augmentations of a single input, we introduce a mean teacher into the PLE scheme. This method helps to reduce noise in the weight maps and stabilize its generation process. Our extensive experimental results on public atrial and prostate segmentation datasets demonstrate that our proposed method achieves state-of-the-art results under semi-supervision. Our code is available at https://github.com/aijinrjinr/MLB-Seg.

Zhuowan Li, Cihang Xie, Benjamin Van Durme et al.

Despite the impressive advancements achieved through vision-and-language pretraining, it remains unclear whether this joint learning paradigm can help understand each individual modality. In this work, we conduct a comparative analysis of the visual representations in existing vision-and-language models and vision-only models by probing a broad range of tasks, aiming to assess the quality of the learned representations in a nuanced manner. Interestingly, our empirical observations suggest that vision-and-language models are better at label prediction tasks like object and attribute prediction, while vision-only models are stronger at dense prediction tasks that require more localized information. We hope our study sheds light on the role of language in visual learning, and serves as an empirical guide for various pretrained models. Code will be released at https://github.com/Lizw14/visual_probing

Zichao Li, Li Liu, Zeyu Wang et al.

Adversarial training (AT) with samples generated by Fast Gradient Sign Method (FGSM), also known as FGSM-AT, is a computationally simple method to train robust networks. However, during its training procedure, an unstable mode of "catastrophic overfitting" has been identified in arXiv:2001.03994 [cs.LG], where the robust accuracy abruptly drops to zero within a single training step. Existing methods use gradient regularizers or random initialization tricks to attenuate this issue, whereas they either take high computational cost or lead to lower robust accuracy. In this work, we provide the first study, which thoroughly examines a collection of tricks from three perspectives: Data Initialization, Network Structure, and Optimization, to overcome the catastrophic overfitting in FGSM-AT. Surprisingly, we find that simple tricks, i.e., a) masking partial pixels (even without randomness), b) setting a large convolution stride and smooth activation functions, or c) regularizing the weights of the first convolutional layer, can effectively tackle the overfitting issue. Extensive results on a range of network architectures validate the effectiveness of each proposed trick, and the combinations of tricks are also investigated. For example, trained with PreActResNet-18 on CIFAR-10, our method attains 49.8% accuracy against PGD-50 attacker and 46.4% accuracy against AutoAttack, demonstrating that pure FGSM-AT is capable of enabling robust learners. The code and models are publicly available at https://github.com/UCSC-VLAA/Bag-of-Tricks-for-FGSM-AT.

Peiran Xu, Zeyu Wang, Jieru Mei et al.

Federated learning (FL) is an emerging paradigm in machine learning, where a shared model is collaboratively learned using data from multiple devices to mitigate the risk of data leakage. While recent studies posit that Vision Transformer (ViT) outperforms Convolutional Neural Networks (CNNs) in addressing data heterogeneity in FL, the specific architectural components that underpin this advantage have yet to be elucidated. In this paper, we systematically investigate the impact of different architectural elements, such as activation functions and normalization layers, on the performance within heterogeneous FL. Through rigorous empirical analyses, we are able to offer the first-of-its-kind general guidance on micro-architecture design principles for heterogeneous FL. Intriguingly, our findings indicate that with strategic architectural modifications, pure CNNs can achieve a level of robustness that either matches or even exceeds that of ViTs when handling heterogeneous data clients in FL. Additionally, our approach is compatible with existing FL techniques and delivers state-of-the-art solutions across a broad spectrum of FL benchmarks. The code is publicly available at https://github.com/UCSC-VLAA/FedConv

Haoqin Tu, Bingchen Zhao, Chen Wei et al.

Multi-modal large language models (MLLMs) are trained based on large language models (LLM), with an enhanced capability to comprehend multi-modal inputs and generate textual responses. While they excel in multi-modal tasks, the pure NLP abilities of MLLMs are often underestimated and left untested. In this study, we get out of the box and unveil an intriguing characteristic of MLLMs -- our preliminary results suggest that visual instruction tuning, a prevailing strategy for transitioning LLMs into MLLMs, unexpectedly and interestingly helps models attain both improved truthfulness and ethical alignment in the pure NLP context. For example, a visual-instruction-tuned LLaMA2 7B model surpasses the performance of the LLaMA2-chat 7B model, fine-tuned with over one million human annotations, on TruthfulQA-mc and Ethics benchmarks. Further analysis reveals that the improved alignment can be attributed to the superior instruction quality inherent to visual-text data. In releasing our code at github.com/UCSC-VLAA/Sight-Beyond-Text, we aspire to foster further exploration into the intrinsic value of visual-text synergies and, in a broader scope, multi-modal interactions in alignment research.

Yuanze Lin, Chen Wei, Huiyu Wang et al.

Video-language pre-training is crucial for learning powerful multi-modal representation. However, it typically requires a massive amount of computation. In this paper, we develop SMAUG, an efficient pre-training framework for video-language models. The foundation component in SMAUG is masked autoencoders. Different from prior works which only mask textual inputs, our masking strategy considers both visual and textual modalities, providing a better cross-modal alignment and saving more pre-training costs. On top of that, we introduce a space-time token sparsification module, which leverages context information to further select only "important" spatial regions and temporal frames for pre-training. Coupling all these designs allows our method to enjoy both competitive performances on text-to-video retrieval and video question answering tasks, and much less pre-training costs by 1.9X or more. For example, our SMAUG only needs about 50 NVIDIA A6000 GPU hours for pre-training to attain competitive performances on these two video-language tasks across six popular benchmarks.

Chen Wang, Angtian Wang, Junbo Li et al.

Neural Radiance Field (NeRF) has demonstrated excellent quality in novel view synthesis, thanks to its ability to model 3D object geometries in a concise formulation. However, current approaches to NeRF-based models rely on clean images with accurate camera calibration, which can be difficult to obtain in the real world, where data is often subject to corruption and distortion. In this work, we provide the first comprehensive analysis of the robustness of NeRF-based novel view synthesis algorithms in the presence of different types of corruptions. We find that NeRF-based models are significantly degraded in the presence of corruption, and are more sensitive to a different set of corruptions than image recognition models. Furthermore, we analyze the robustness of the feature encoder in generalizable methods, which synthesize images using neural features extracted via convolutional neural networks or transformers, and find that it only contributes marginally to robustness. Finally, we reveal that standard data augmentation techniques, which can significantly improve the robustness of recognition models, do not help the robustness of NeRF-based models. We hope that our findings will attract more researchers to study the robustness of NeRF-based approaches and help to improve their performance in the real world.

Yunfei Xie, Ce Zhou, Lang Gao et al.

This paper introduces MedTrinity-25M, a comprehensive, large-scale multimodal dataset for medicine, covering over 25 million images across 10 modalities with multigranular annotations for more than 65 diseases. These multigranular annotations encompass both global information, such as modality and organ detection, and local information like ROI analysis, lesion texture, and region-wise correlations. Unlike the existing multimodal datasets, which are limited by the availability of image-text pairs, we have developed the first automated pipeline that scales up multimodal data by generating multigranular visual and textual annotations in the form of image-ROI-description triplets without the need for any paired text descriptions. Specifically, data from over 30 different sources have been collected, preprocessed, and grounded using domain-specific expert models to identify ROIs related to abnormal regions. We then build a comprehensive knowledge base and prompt multimodal large language models to perform retrieval-augmented generation with the identified ROIs as guidance, resulting in multigranular textual descriptions. Compared to existing datasets, MedTrinity-25M provides the most enriched annotations, supporting a comprehensive range of multimodal tasks such as captioning and report generation, as well as vision-centric tasks like classification and segmentation. We propose LLaVA-Tri by pretraining LLaVA on MedTrinity-25M, achieving state-of-the-art performance on VQA-RAD, SLAKE, and PathVQA, surpassing representative SOTA multimodal large language models. Furthermore, MedTrinity-25M can also be utilized to support large-scale pre-training of multimodal medical AI models, contributing to the development of future foundation models in the medical domain. We will make our dataset available.

Nataniel Ruiz, Sarah Adel Bargal, Cihang Xie et al.

Modern deep neural networks tend to be evaluated on static test sets. One shortcoming of this is the fact that these deep neural networks cannot be easily evaluated for robustness issues with respect to specific scene variations. For example, it is hard to study the robustness of these networks to variations of object scale, object pose, scene lighting and 3D occlusions. The main reason is that collecting real datasets with fine-grained naturalistic variations of sufficient scale can be extremely time-consuming and expensive. In this work, we present Counterfactual Simulation Testing, a counterfactual framework that allows us to study the robustness of neural networks with respect to some of these naturalistic variations by building realistic synthetic scenes that allow us to ask counterfactual questions to the models, ultimately providing answers to questions such as "Would your classification still be correct if the object were viewed from the top?" or "Would your classification still be correct if the object were partially occluded by another object?". Our method allows for a fair comparison of the robustness of recently released, state-of-the-art Convolutional Neural Networks and Vision Transformers, with respect to these naturalistic variations. We find evidence that ConvNext is more robust to pose and scale variations than Swin, that ConvNext generalizes better to our simulated domain and that Swin handles partial occlusion better than ConvNext. We also find that robustness for all networks improves with network scale and with data scale and variety. We release the Naturalistic Variation Object Dataset (NVD), a large simulated dataset of 272k images of everyday objects with naturalistic variations such as object pose, scale, viewpoint, lighting and occlusions. Project page: https://counterfactualsimulation.github.io

Zeyu Wang, Dingwen Li, Chenxu Luo et al.

3D perception based on the representations learned from multi-camera bird's-eye-view (BEV) is trending as cameras are cost-effective for mass production in autonomous driving industry. However, there exists a distinct performance gap between multi-camera BEV and LiDAR based 3D object detection. One key reason is that LiDAR captures accurate depth and other geometry measurements, while it is notoriously challenging to infer such 3D information from merely image input. In this work, we propose to boost the representation learning of a multi-camera BEV based student detector by training it to imitate the features of a well-trained LiDAR based teacher detector. We propose effective balancing strategy to enforce the student to focus on learning the crucial features from the teacher, and generalize knowledge transfer to multi-scale layers with temporal fusion. We conduct extensive evaluations on multiple representative models of multi-camera BEV. Experiments reveal that our approach renders significant improvement over the student models, leading to the state-of-the-art performance on the popular benchmark nuScenes.

Jiaqi Liu, Yaofeng Su, Peng Xia et al.

To support long-term interaction in complex environments, LLM agents require memory systems that manage historical experiences. Existing approaches either retain full interaction histories via passive context extension, leading to substantial redundancy, or rely on iterative reasoning to filter noise, incurring high token costs. To address this challenge, we introduce SimpleMem, an efficient memory framework based on semantic lossless compression. We propose a three-stage pipeline designed to maximize information density and token utilization: (1) Semantic Structured Compression, which distills unstructured interactions into compact, multi-view indexed memory units; (2) Online Semantic Synthesis, an intra-session process that instantly integrates related context into unified abstract representations to eliminate redundancy; and (3) Intent-Aware Retrieval Planning, which infers search intent to dynamically determine retrieval scope and construct precise context efficiently. Experiments on benchmark datasets show that our method consistently outperforms baseline approaches in accuracy, retrieval efficiency, and inference cost, achieving an average F1 improvement of 26.4% in LoCoMo while reducing inference-time token consumption by up to 30-fold, demonstrating a superior balance between performance and efficiency. Code is available at https://github.com/aiming-lab/SimpleMem.

Shutong Wu, Sizhe Chen, Cihang Xie et al.

Unlearnable examples (ULEs) aim to protect data from unauthorized usage for training DNNs. Existing work adds $\ell_\infty$-bounded perturbations to the original sample so that the trained model generalizes poorly. Such perturbations, however, are easy to eliminate by adversarial training and data augmentations. In this paper, we resolve this problem from a novel perspective by perturbing only one pixel in each image. Interestingly, such a small modification could effectively degrade model accuracy to almost an untrained counterpart. Moreover, our produced \emph{One-Pixel Shortcut (OPS)} could not be erased by adversarial training and strong augmentations. To generate OPS, we perturb in-class images at the same position to the same target value that could mostly and stably deviate from all the original images. Since such generation is only based on images, OPS needs significantly less computation cost than the previous methods using DNN generators. Based on OPS, we introduce an unlearnable dataset called CIFAR-10-S, which is indistinguishable from CIFAR-10 by humans but induces the trained model to extremely low accuracy. Even under adversarial training, a ResNet-18 trained on CIFAR-10-S has only 10.61% accuracy, compared to 83.02% by the existing error-minimizing method.

Shaoyuan Xie, Zichao Li, Zeyu Wang et al.

In recent years, camera-based 3D object detection has gained widespread attention for its ability to achieve high performance with low computational cost. However, the robustness of these methods to adversarial attacks has not been thoroughly examined, especially when considering their deployment in safety-critical domains like autonomous driving. In this study, we conduct the first comprehensive investigation of the robustness of leading camera-based 3D object detection approaches under various adversarial conditions. We systematically analyze the resilience of these models under two attack settings: white-box and black-box; focusing on two primary objectives: classification and localization. Additionally, we delve into two types of adversarial attack techniques: pixel-based and patch-based. Our experiments yield four interesting findings: (a) bird's-eye-view-based representations exhibit stronger robustness against localization attacks; (b) depth-estimation-free approaches have the potential to show stronger robustness; (c) accurate depth estimation effectively improves robustness for depth-estimation-based methods; (d) incorporating multi-frame benign inputs can effectively mitigate adversarial attacks. We hope our findings can steer the development of future camera-based object detection models with enhanced adversarial robustness.

Yunchao Zhang, Zonglin Di, Kaiwen Zhou et al.

Federated embodied agent learning protects the data privacy of individual visual environments by keeping data locally at each client (the individual environment) during training. However, since the local data is inaccessible to the server under federated learning, attackers may easily poison the training data of the local client to build a backdoor in the agent without notice. Deploying such an agent raises the risk of potential harm to humans, as the attackers may easily navigate and control the agent as they wish via the backdoor. Towards Byzantine-robust federated embodied agent learning, in this paper, we study the attack and defense for the task of vision-and-language navigation (VLN), where the agent is required to follow natural language instructions to navigate indoor environments. First, we introduce a simple but effective attack strategy, Navigation as Wish (NAW), in which the malicious client manipulates local trajectory data to implant a backdoor into the global model. Results on two VLN datasets (R2R and RxR) show that NAW can easily navigate the deployed VLN agent regardless of the language instruction, without affecting its performance on normal test sets. Then, we propose a new Prompt-Based Aggregation (PBA) to defend against the NAW attack in federated VLN, which provides the server with a ''prompt'' of the vision-and-language alignment variance between the benign and malicious clients so that they can be distinguished during training. We validate the effectiveness of the PBA method on protecting the global model from the NAW attack, which outperforms other state-of-the-art defense methods by a large margin in the defense metrics on R2R and RxR.

Yunfei Xie, Cihang Xie, Alan Yuille et al.

In this paper, we introduce a hierarchical transformer-based model designed for sophisticated image segmentation tasks, effectively bridging the granularity of part segmentation with the comprehensive scope of object segmentation. At the heart of our approach is a multi-level representation strategy, which systematically advances from individual pixels to superpixels, and ultimately to cohesive group formations. This architecture is underpinned by two pivotal aggregation strategies: local aggregation and global aggregation. Local aggregation is employed to form superpixels, leveraging the inherent redundancy of the image data to produce segments closely aligned with specific parts of the object, guided by object-level supervision. In contrast, global aggregation interlinks these superpixels, organizing them into larger groups that correlate with entire objects and benefit from part-level supervision. This dual aggregation framework ensures a versatile adaptation to varying supervision inputs while maintaining computational efficiency. Our methodology notably improves the balance between adaptability across different supervision modalities and computational manageability, culminating in significant enhancement in segmentation performance. When tested on the PartImageNet dataset, our model achieves a substantial increase, outperforming the previous state-of-the-art by 2.8% and 0.8% in mIoU scores for part and object segmentation, respectively. Similarly, on the Pascal Part dataset, it records performance enhancements of 1.5% and 2.0% for part and object segmentation, respectively.

Hardy Chen, Haoqin Tu, Fali Wang et al.

This work revisits the dominant supervised fine-tuning (SFT) then reinforcement learning (RL) paradigm for training Large Vision-Language Models (LVLMs), and reveals a key finding: SFT can significantly undermine subsequent RL by inducing ``pseudo reasoning paths'' imitated from expert models. While these paths may resemble the native reasoning paths of RL models, they often involve prolonged, hesitant, less informative steps, and incorrect reasoning. To systematically study this effect, we introduce VLAA-Thinking, a new multimodal dataset designed to support reasoning in LVLMs. Constructed via a six-step pipeline involving captioning, reasoning distillation, answer rewrite and verification, VLAA-Thinking comprises high-quality, step-by-step visual reasoning traces for SFT, along with a more challenging RL split from the same data source. Using this dataset, we conduct extensive experiments comparing SFT, RL and their combinations. Results show that while SFT helps models learn reasoning formats, it often locks aligned models into imitative, rigid reasoning modes that impede further learning. In contrast, building on the Group Relative Policy Optimization (GRPO) with a novel mixed reward module integrating both perception and cognition signals, our RL approach fosters more genuine, adaptive reasoning behavior. Notably, our model VLAA-Thinker, based on Qwen2.5VL 3B, achieves top-1 performance on Open LMM Reasoning Leaderboard (https://huggingface.co/spaces/opencompass/Open_LMM_Reasoning_Leaderboard) among 4B scale LVLMs, surpassing the previous state-of-the-art by 1.8%. We hope our findings provide valuable insights in developing reasoning-capable LVLMs and can inform future research in this area.

Feng Wang, Jiahao Wang, Sucheng Ren et al.

Similar to Vision Transformers, this paper identifies artifacts also present within the feature maps of Vision Mamba. These artifacts, corresponding to high-norm tokens emerging in low-information background areas of images, appear much more severe in Vision Mamba -- they exist prevalently even with the tiny-sized model and activate extensively across background regions. To mitigate this issue, we follow the prior solution of introducing register tokens into Vision Mamba. To better cope with Mamba blocks' uni-directional inference paradigm, two key modifications are introduced: 1) evenly inserting registers throughout the input token sequence, and 2) recycling registers for final decision predictions. We term this new architecture Mamba-R. Qualitative observations suggest, compared to vanilla Vision Mamba, Mamba-R's feature maps appear cleaner and more focused on semantically meaningful regions. Quantitatively, Mamba-R attains stronger performance and scales better. For example, on the ImageNet benchmark, our base-size Mamba-R attains 83.0% accuracy, significantly outperforming Vim-B's 81.8%; furthermore, we provide the first successful scaling to the large model size (i.e., with 341M parameters), attaining a competitive accuracy of 83.6% (84.5% if finetuned with 384x384 inputs). Additional validation on the downstream semantic segmentation task also supports Mamba-R's efficacy. Code is available at https://github.com/wangf3014/Mamba-Reg.

Zijun Wang, Haoqin Tu, Yuhan Wang et al.

This paper introduces STAR-1, a high-quality, just-1k-scale safety dataset specifically designed for large reasoning models (LRMs) like DeepSeek-R1. Built on three core principles -- diversity, deliberative reasoning, and rigorous filtering -- STAR-1 aims to address the critical needs for safety alignment in LRMs. Specifically, we begin by integrating existing open-source safety datasets from diverse sources. Then, we curate safety policies to generate policy-grounded deliberative reasoning samples. Lastly, we apply a GPT-4o-based safety scoring system to select training examples aligned with best practices. Experimental results show that fine-tuning LRMs with STAR-1 leads to an average 40% improvement in safety performance across four benchmarks, while only incurring a marginal decrease (e.g., an average of 1.1%) in reasoning ability measured across five reasoning tasks. Extensive ablation studies further validate the importance of our design principles in constructing STAR-1 and analyze its efficacy across both LRMs and traditional LLMs. Our project page is https://ucsc-vlaa.github.io/STAR-1.

Zeyu Wang, Xianhang Li, Hongru Zhu et al.

The machine learning community has witnessed a drastic change in the training pipeline, pivoted by those ''foundation models'' with unprecedented scales. However, the field of adversarial training is lagging behind, predominantly centered around small model sizes like ResNet-50, and tiny and low-resolution datasets like CIFAR-10. To bridge this transformation gap, this paper provides a modern re-examination with adversarial training, investigating its potential benefits when applied at scale. Additionally, we introduce an efficient and effective training strategy to enable adversarial training with giant models and web-scale data at an affordable computing cost. We denote this newly introduced framework as AdvXL. Empirical results demonstrate that AdvXL establishes new state-of-the-art robust accuracy records under AutoAttack on ImageNet-1K. For example, by training on DataComp-1B dataset, our AdvXL empowers a vanilla ViT-g model to substantially surpass the previous records of $l_{\infty}$-, $l_{2}$-, and $l_{1}$-robust accuracy by margins of 11.4%, 14.2% and 12.9%, respectively. This achievement posits AdvXL as a pioneering approach, charting a new trajectory for the efficient training of robust visual representations at significantly larger scales. Our code is available at https://github.com/UCSC-VLAA/AdvXL.

Yuhan Wang, Siwei Yang, Bingchen Zhao et al.

Recent advancements in large multimodal models like GPT-4o have set a new standard for high-fidelity, instruction-guided image editing. However, the proprietary nature of these models and their training data creates a significant barrier for open-source research. To bridge this gap, we introduce GPT-IMAGE-EDIT-1.5M, a publicly available, large-scale image-editing corpus containing more than 1.5 million high-quality triplets (instruction, source image, edited image). We systematically construct this dataset by leveraging the versatile capabilities of GPT-4o to unify and refine three popular image-editing datasets: OmniEdit, HQ-Edit, and UltraEdit. Specifically, our methodology involves 1) regenerating output images to enhance visual quality and instruction alignment, and 2) selectively rewriting prompts to improve semantic clarity. To validate the efficacy of our dataset, we fine-tune advanced open-source models on GPT-IMAGE-EDIT-1.5M. The empirical results are exciting, e.g., the fine-tuned FluxKontext achieves highly competitive performance across a comprehensive suite of benchmarks, including 7.24 on GEdit-EN, 3.80 on ImgEdit-Full, and 8.78 on Complex-Edit, showing stronger instruction following and higher perceptual quality while maintaining identity. These scores markedly exceed all previously published open-source methods and substantially narrow the gap to leading proprietary models. We hope the full release of GPT-IMAGE-EDIT-1.5M can help to catalyze further open research in instruction-guided image editing.

Zijun Wang, Haoqin Tu, Jieru Mei et al.

This paper studies the vulnerabilities of transformer-based Large Language Models (LLMs) to jailbreaking attacks, focusing specifically on the optimization-based Greedy Coordinate Gradient (GCG) strategy. We first observe a positive correlation between the effectiveness of attacks and the internal behaviors of the models. For instance, attacks tend to be less effective when models pay more attention to system prompts designed to ensure LLM safety alignment. Building on this discovery, we introduce an enhanced method that manipulates models' attention scores to facilitate LLM jailbreaking, which we term AttnGCG. Empirically, AttnGCG shows consistent improvements in attack efficacy across diverse LLMs, achieving an average increase of ~7% in the Llama-2 series and ~10% in the Gemma series. Our strategy also demonstrates robust attack transferability against both unseen harmful goals and black-box LLMs like GPT-3.5 and GPT-4. Moreover, we note our attention-score visualization is more interpretable, allowing us to gain better insights into how our targeted attention manipulation facilitates more effective jailbreaking. We release the code at https://github.com/UCSC-VLAA/AttnGCG-attack.

Feng Wang, Yaodong Yu, Guoyizhe Wei et al.

Since the introduction of Vision Transformer (ViT), patchification has long been regarded as a de facto image tokenization approach for plain visual architectures. By compressing the spatial size of images, this approach can effectively shorten the token sequence and reduce the computational cost of ViT-like plain architectures. In this work, we aim to thoroughly examine the information loss caused by this patchification-based compressive encoding paradigm and how it affects visual understanding. We conduct extensive patch size scaling experiments and excitedly observe an intriguing scaling law in patchification: the models can consistently benefit from decreased patch sizes and attain improved predictive performance, until it reaches the minimum patch size of 1x1, i.e., pixel tokenization. This conclusion is broadly applicable across different vision tasks, various input scales, and diverse architectures such as ViT and the recent Mamba models. Moreover, as a by-product, we discover that with smaller patches, task-specific decoder heads become less critical for dense prediction. In the experiments, we successfully scale up the visual sequence to an exceptional length of 50,176 tokens, achieving a competitive test accuracy of 84.6% with a base-sized model on the ImageNet-1k benchmark. We hope this study can provide insights and theoretical foundations for future works of building non-compressive vision models. Code is available at https://github.com/wangf3014/Patch_Scaling.

Sucheng Ren, Zeyu Wang, Hongru Zhu et al.

This paper enhances image-GPT (iGPT), one of the pioneering works that introduce autoregressive pretraining to predict the next pixels for visual representation learning. Two simple yet essential changes are made. First, we shift the prediction target from raw pixels to semantic tokens, enabling a higher-level understanding of visual content. Second, we supplement the autoregressive modeling by instructing the model to predict not only the next tokens but also the visible tokens. This pipeline is particularly effective when semantic tokens are encoded by discriminatively trained models, such as CLIP. We introduce this novel approach as D-iGPT. Extensive experiments showcase that D-iGPT excels as a strong learner of visual representations: A notable achievement is its compelling performance on the ImageNet-1K dataset -- by training on publicly available datasets, D-iGPT unprecedentedly achieves \textbf{90.0\%} top-1 accuracy with a vanilla ViT-H. Additionally, D-iGPT shows strong generalization on the downstream task. Code is available at https://github.com/OliverRensu/D-iGPT.

Haoqin Tu, Weitao Feng, Hardy Chen et al.

Process-supervised reward models serve as a fine-grained function that provides detailed step-wise feedback to model responses, facilitating effective selection of reasoning trajectories for complex tasks. Despite its advantages, evaluation on PRMs remains less explored, especially in the multimodal domain. To address this gap, this paper first benchmarks current vision large language models (VLLMs) as two types of reward models: output reward models (ORMs) and process reward models (PRMs) on multiple vision-language benchmarks, which reveal that neither ORM nor PRM consistently outperforms across all tasks, and superior VLLMs do not necessarily yield better rewarding performance. To further advance evaluation, we introduce ViLBench, a vision-language benchmark designed to require intensive process reward signals. Notably, OpenAI's GPT-4o with Chain-of-Thought (CoT) achieves only 27.3% accuracy, indicating the benchmark's challenge for current VLLMs. Lastly, we preliminarily showcase a promising pathway towards bridging the gap between general VLLMs and reward models -- by collecting 73.6K vision-language process reward data using an enhanced tree-search algorithm, our 3B model is able to achieve an average improvement of 3.3% over standard CoT and up to 2.5% compared to its untrained counterpart on ViLBench by selecting OpenAI o1's generations. We release the implementations at https://ucsc-vlaa.github.io/ViLBench with our code, model, and data.

Sucheng Ren, Yaodong Yu, Nataniel Ruiz et al.

There exists recent work in computer vision, named VAR, that proposes a new autoregressive paradigm for image generation. Diverging from the vanilla next-token prediction, VAR structurally reformulates the image generation into a coarse to fine next-scale prediction. In this paper, we show that this scale-wise autoregressive framework can be effectively decoupled into \textit{intra-scale modeling}, which captures local spatial dependencies within each scale, and \textit{inter-scale modeling}, which models cross-scale relationships progressively from coarse-to-fine scales. This decoupling structure allows to rebuild VAR in a more computationally efficient manner. Specifically, for intra-scale modeling -- crucial for generating high-fidelity images -- we retain the original bidirectional self-attention design to ensure comprehensive modeling; for inter-scale modeling, which semantically connects different scales but is computationally intensive, we apply linear-complexity mechanisms like Mamba to substantially reduce computational overhead. We term this new framework M-VAR. Extensive experiments demonstrate that our method outperforms existing models in both image quality and generation speed. For example, our 1.5B model, with fewer parameters and faster inference speed, outperforms the largest VAR-d30-2B. Moreover, our largest model M-VAR-d32 impressively registers 1.78 FID on ImageNet 256$\times$256 and outperforms the prior-art autoregressive models LlamaGen/VAR by 0.4/0.19 and popular diffusion models LDM/DiT by 1.82/0.49, respectively. Code is avaiable at \url{https://github.com/OliverRensu/MVAR}.

Siwei Yang, Mude Hui, Bingchen Zhao et al.

We introduce $\texttt{Complex-Edit}$, a comprehensive benchmark designed to systematically evaluate instruction-based image editing models across instructions of varying complexity. To develop this benchmark, we harness GPT-4o to automatically collect a diverse set of editing instructions at scale. Our approach follows a well-structured ``Chain-of-Edit'' pipeline: we first generate individual atomic editing tasks independently and then integrate them to form cohesive, complex instructions. Additionally, we introduce a suite of metrics to assess various aspects of editing performance, along with a VLM-based auto-evaluation pipeline that supports large-scale assessments. Our benchmark yields several notable insights: 1) Open-source models significantly underperform relative to proprietary, closed-source models, with the performance gap widening as instruction complexity increases; 2) Increased instructional complexity primarily impairs the models' ability to retain key elements from the input images and to preserve the overall aesthetic quality; 3) Decomposing a complex instruction into a sequence of atomic steps, executed in a step-by-step manner, substantially degrades performance across multiple metrics; 4) A straightforward Best-of-N selection strategy improves results for both direct editing and the step-by-step sequential approach; and 5) We observe a ``curse of synthetic data'': when synthetic data is involved in model training, the edited images from such models tend to appear increasingly synthetic as the complexity of the editing instructions rises -- a phenomenon that intriguingly also manifests in the latest GPT-4o outputs.

Siwei Yang, Bingchen Zhao, Cihang Xie

This paper introduces AQA-Bench, a novel benchmark to assess the sequential reasoning capabilities of large language models (LLMs) in algorithmic contexts, such as depth-first search (DFS). The key feature of our evaluation benchmark lies in its interactive evaluation protocol - for example, in DFS, the availability of each node's connected edge is contingent upon the model's traversal to that node, thereby necessitating the LLM's ability to effectively remember visited nodes and strategize subsequent moves considering the possible environmental feedback in the future steps. We comprehensively build AQA-Bench with three different algorithms, namely binary search, depth-first search, and breadth-first search, and to evaluate the sequential reasoning ability of 14 different LLMs. Our investigations reveal several interesting findings: (1) Closed-source models like GPT-4 and Gemini generally show much stronger sequential reasoning ability, significantly outperforming open-source LLMs. (2) Naively providing in-context examples may inadvertently hurt few-shot performance in an interactive environment due to over-fitting to examples. (3) Instead of using optimal steps from another test case as the in-context example, a very limited number of predecessor steps in the current test case following the optimal policy can substantially boost small models' performance. (4) The performance gap between weak models and strong models is greatly due to the incapability of weak models to start well. (5) The scaling correlation between performance and model size is not always significant, sometimes even showcasing an inverse trend. We hope our study can catalyze future work on advancing the understanding and enhancement of LLMs' capabilities in sequential reasoning. The code is available at https://github.com/UCSC-VLAA/AQA-Bench.

Siwei Han, Jiaqi Liu, Yaofeng Su et al.

As Large Language Model (LLM) agents increasingly gain self-evolutionary capabilities to adapt and refine their strategies through real-world interaction, their long-term reliability becomes a critical concern. We identify the Alignment Tipping Process (ATP), a critical post-deployment risk unique to self-evolving LLM agents. Unlike training-time failures, ATP arises when continual interaction drives agents to abandon alignment constraints established during training in favor of reinforced, self-interested strategies. We formalize and analyze ATP through two complementary paradigms: Self-Interested Exploration, where repeated high-reward deviations induce individual behavioral drift, and Imitative Strategy Diffusion, where deviant behaviors spread across multi-agent systems. Building on these paradigms, we construct controllable testbeds and benchmark Qwen3-8B and Llama-3.1-8B-Instruct. Our experiments show that alignment benefits erode rapidly under self-evolution, with initially aligned models converging toward unaligned states. In multi-agent settings, successful violations diffuse quickly, leading to collective misalignment. Moreover, current reinforcement learning-based alignment methods provide only fragile defenses against alignment tipping. Together, these findings demonstrate that alignment of LLM agents is not a static property but a fragile and dynamic one, vulnerable to feedback-driven decay during deployment. Our data and code are available at https://github.com/aiming-lab/ATP.

Junfei Xiao, Ziqi Zhou, Wenxuan Li et al.

This paper introduces ProLab, a novel approach using property-level label space for creating strong interpretable segmentation models. Instead of relying solely on category-specific annotations, ProLab uses descriptive properties grounded in common sense knowledge for supervising segmentation models. It is based on two core designs. First, we employ Large Language Models (LLMs) and carefully crafted prompts to generate descriptions of all involved categories that carry meaningful common sense knowledge and follow a structured format. Second, we introduce a description embedding model preserving semantic correlation across descriptions and then cluster them into a set of descriptive properties (e.g., 256) using K-Means. These properties are based on interpretable common sense knowledge consistent with theories of human recognition. We empirically show that our approach makes segmentation models perform stronger on five classic benchmarks (e.g., ADE20K, COCO-Stuff, Pascal Context, Cityscapes, and BDD). Our method also shows better scalability with extended training steps than category-level supervision. Our interpretable segmentation framework also emerges with the generalization ability to segment out-of-domain or unknown categories using only in-domain descriptive properties. Code is available at https://github.com/lambert-x/ProLab.

Mude Hui, Siwei Yang, Bingchen Zhao et al.

This study introduces HQ-Edit, a high-quality instruction-based image editing dataset with around 200,000 edits. Unlike prior approaches relying on attribute guidance or human feedback on building datasets, we devise a scalable data collection pipeline leveraging advanced foundation models, namely GPT-4V and DALL-E 3. To ensure its high quality, diverse examples are first collected online, expanded, and then used to create high-quality diptychs featuring input and output images with detailed text prompts, followed by precise alignment ensured through post-processing. In addition, we propose two evaluation metrics, Alignment and Coherence, to quantitatively assess the quality of image edit pairs using GPT-4V. HQ-Edits high-resolution images, rich in detail and accompanied by comprehensive editing prompts, substantially enhance the capabilities of existing image editing models. For example, an HQ-Edit finetuned InstructPix2Pix can attain state-of-the-art image editing performance, even surpassing those models fine-tuned with human-annotated data. The project page is https://thefllood.github.io/HQEdit_web.

Xianhang Li, Haoqin Tu, Mude Hui et al.

Web-crawled image-text pairs are inherently noisy. Prior studies demonstrate that semantically aligning and enriching textual descriptions of these pairs can significantly enhance model training across various vision-language tasks, particularly text-to-image generation. However, large-scale investigations in this area remain predominantly closed-source. Our paper aims to bridge this community effort, leveraging the powerful and \textit{open-sourced} LLaMA-3, a GPT-4 level LLM. Our recaptioning pipeline is simple: first, we fine-tune a LLaMA-3-8B powered LLaVA-1.5 and then employ it to recaption 1.3 billion images from the DataComp-1B dataset. Our empirical results confirm that this enhanced dataset, Recap-DataComp-1B, offers substantial benefits in training advanced vision-language models. For discriminative models like CLIP, we observe enhanced zero-shot performance in cross-modal retrieval tasks. For generative models like text-to-image Diffusion Transformers, the generated images exhibit a significant improvement in alignment with users' text instructions, especially in following complex queries. Our project page is https://www.haqtu.me/Recap-Datacomp-1B/

Sucheng Ren, Xianhang Li, Haoqin Tu et al.

The vision community has started to build with the recently developed state space model, Mamba, as the new backbone for a range of tasks. This paper shows that Mamba's visual capability can be significantly enhanced through autoregressive pretraining, a direction not previously explored. Efficiency-wise, the autoregressive nature can well capitalize on the Mamba's unidirectional recurrent structure, enabling faster overall training speed compared to other training strategies like mask modeling. Performance-wise, autoregressive pretraining equips the Mamba architecture with markedly higher accuracy over its supervised-trained counterparts and, more importantly, successfully unlocks its scaling potential to large and even huge model sizes. For example, with autoregressive pretraining, a base-size Mamba attains 83.2\% ImageNet accuracy, outperforming its supervised counterpart by 2.0\%; our huge-size Mamba, the largest Vision Mamba to date, attains 85.0\% ImageNet accuracy (85.5\% when finetuned with $384\times384$ inputs), notably surpassing all other Mamba variants in vision. The code is available at \url{https://github.com/OliverRensu/ARM}.

Xianhang Li, Zeyu Wang, Cihang Xie

CLIP, one of the pioneering foundation models that connect images and text, has enabled many recent breakthroughs in computer vision. However, its associated training cost is prohibitively high, imposing a significant barrier to its widespread exploration. In this paper, we present a surprising finding that there exists an inverse scaling law for CLIP training, whereby the larger the image/text encoders used, the shorter the sequence length of image/text tokens that can be applied in training. Moreover, we showcase that the strategy for reducing image/text token length plays a crucial role in determining the quality of this scaling law. As a result of this finding, we are able to successfully train CLIP even with limited computational resources. For example, using 8 A100 GPUs, our CLIP models achieve zero-shot top-1 ImageNet-1k accuracies of 63.2% in ~2 days, 67.8% in ~3 days, and 69.3% in ~4 days. Our method also works well when scaling up -- with G/14, we register a new record of 83.0% ImageNet-1k zero-shot accuracy, and meanwhile accelerate the training by ~33x compared to its OpenCLIP counterpart. By reducing the computation barrier associated with CLIP, we hope to inspire more research in this field, particularly from academics. Our code is available at https://github.com/UCSC-VLAA/CLIPA.

Yuyin Zhou, Xianhang Li, Fengze Liu et al.

Deep neural networks have shown great success in representation learning. However, when learning with noisy labels (LNL), they can easily overfit and fail to generalize to new data. This paper introduces a simple and effective method, named Learning to Bootstrap (L2B), which enables models to bootstrap themselves using their own predictions without being adversely affected by erroneous pseudo-labels. It achieves this by dynamically adjusting the importance weight between real observed and generated labels, as well as between different samples through meta-learning. Unlike existing instance reweighting methods, the key to our method lies in a new, versatile objective that enables implicit relabeling concurrently, leading to significant improvements without incurring additional costs. L2B offers several benefits over the baseline methods. It yields more robust models that are less susceptible to the impact of noisy labels by guiding the bootstrapping procedure more effectively. It better exploits the valuable information contained in corrupted instances by adapting the weights of both instances and labels. Furthermore, L2B is compatible with existing LNL methods and delivers competitive results spanning natural and medical imaging tasks including classification and segmentation under both synthetic and real-world noise. Extensive experiments demonstrate that our method effectively mitigates the challenges of noisy labels, often necessitating few to no validation samples, and is well generalized to other tasks such as image segmentation. This not only positions it as a robust complement to existing LNL techniques but also underscores its practical applicability. The code and models are available at https://github.com/yuyinzhou/l2b.

Yutong Bai, Jieru Mei, Alan Yuille et al.

Transformer emerges as a powerful tool for visual recognition. In addition to demonstrating competitive performance on a broad range of visual benchmarks, recent works also argue that Transformers are much more robust than Convolutions Neural Networks (CNNs). Nonetheless, surprisingly, we find these conclusions are drawn from unfair experimental settings, where Transformers and CNNs are compared at different scales and are applied with distinct training frameworks. In this paper, we aim to provide the first fair & in-depth comparisons between Transformers and CNNs, focusing on robustness evaluations. With our unified training setup, we first challenge the previous belief that Transformers outshine CNNs when measuring adversarial robustness. More surprisingly, we find CNNs can easily be as robust as Transformers on defending against adversarial attacks, if they properly adopt Transformers' training recipes. While regarding generalization on out-of-distribution samples, we show pre-training on (external) large-scale datasets is not a fundamental request for enabling Transformers to achieve better performance than CNNs. Moreover, our ablations suggest such stronger generalization is largely benefited by the Transformer's self-attention-like architectures per se, rather than by other training setups. We hope this work can help the community better understand and benchmark the robustness of Transformers and CNNs. The code and models are publicly available at https://github.com/ytongbai/ViTs-vs-CNNs.

Zhuowan Li, Elias Stengel-Eskin, Yixiao Zhang et al.

While neural symbolic methods demonstrate impressive performance in visual question answering on synthetic images, their performance suffers on real images. We identify that the long-tail distribution of visual concepts and unequal importance of reasoning steps in real data are the two key obstacles that limit the models' real-world potentials. To address these challenges, we propose a new paradigm, Calibrating Concepts and Operations (CCO), which enables neural symbolic models to capture underlying data characteristics and to reason with hierarchical importance. Specifically, we introduce an executor with learnable concept embedding magnitudes for handling distribution imbalance, and an operation calibrator for highlighting important operations and suppressing redundant ones. Our experiments show CCO substantially boosts the performance of neural symbolic methods on real images. By evaluating models on the real world dataset GQA, CCO helps the neural symbolic method NSCL outperforms its vanilla counterpart by 9.1% (from 47.0% to 56.1%); this result also largely reduces the performance gap between symbolic and non-symbolic methods. Additionally, we create a perturbed test set for better understanding and analyzing model performance on real images. Code is available at https://github.com/Lizw14/CaliCO.git .

Xiangning Chen, Cihang Xie, Mingxing Tan et al.

Data augmentation has become a de facto component for training high-performance deep image classifiers, but its potential is under-explored for object detection. Noting that most state-of-the-art object detectors benefit from fine-tuning a pre-trained classifier, we first study how the classifiers' gains from various data augmentations transfer to object detection. The results are discouraging; the gains diminish after fine-tuning in terms of either accuracy or robustness. This work instead augments the fine-tuning stage for object detectors by exploring adversarial examples, which can be viewed as a model-dependent data augmentation. Our method dynamically selects the stronger adversarial images sourced from a detector's classification and localization branches and evolves with the detector to ensure the augmentation policy stays current and relevant. This model-dependent augmentation generalizes to different object detectors better than AutoAugment, a model-agnostic augmentation policy searched based on one particular detector. Our approach boosts the performance of state-of-the-art EfficientDets by +1.1 mAP on the COCO object detection benchmark. It also improves the detectors' robustness against natural distortions by +3.8 mAP and against domain shift by +1.3 mAP. Models are available at https://github.com/google/automl/tree/master/efficientdet/Det-AdvProp.md