PyTorchKR Papers

“Code Generation”

총 10,186편
태그·기간 등 상세 필터

카테고리 탐색

arXiv:2304.09728공식 코드 · Papers with Code

Any-to-Any Style Transfer: Making Picasso and Da Vinci Collaborate

Style transfer aims to render the style of a given image for style reference to another given image for content reference, and has been widely adopted in artistic generation and image editing. Existing approaches either apply the holistic style of the style image in a global manner, or migrate local colors and textures of the style image to the content counterparts in a pre-defined way. In either case, only one result can be generated for a specific pair of content and style images, which therefore lacks flexibility and is hard to satisfy different users with different preferences. We propose here a novel strategy termed Any-to-Any Style Transfer to address this drawback, which enables users to interactively select styles of regions in the style image and apply them to the prescribed content regions. In this way, personalizable style transfer is achieved through human-computer interaction. At the heart of our approach lies in (1) a region segmentation module based on Segment Anything, which supports region selection with only some clicks or drawing on images and thus takes user inputs conveniently and flexibly; (2) and an attention fusion module, which converts inputs from users to controlling signals for the style transfer model. Experiments demonstrate the effectiveness for personalizable style transfer. Notably, our approach performs in a plug-and-play manner portable to any style transfer method and enhance the controllablity. Our code is available \href{https://github.com/Huage001/Transfer-Any-Style}{here}.

arXiv:2305.00282

NSLF-OL: Online Learning of Neural Surface Light Fields alongside Real-time Incremental 3D Reconstruction

Immersive novel view generation is an important technology in the field of graphics and has recently also received attention for operator-based human-robot interaction. However, the involved training is time-consuming, and thus the current test scope is majorly on object capturing. This limits the usage of related models in the robotics community for 3D reconstruction since robots (1) usually only capture a very small range of view directions to surfaces that cause arbitrary predictions on unseen, novel direction, (2) requires real-time algorithms, and (3) work with growing scenes, e.g., in robotic exploration. The paper proposes a novel Neural Surface Light Fields model that copes with the small range of view directions while producing a good result in unseen directions. Exploiting recent encoding techniques, the training of our model is highly efficient. In addition, we design Multiple Asynchronous Neural Agents (MANA), a universal framework to learn each small region in parallel for large-scale growing scenes. Our model learns online the Neural Surface Light Fields (NSLF) aside from real-time 3D reconstruction with a sequential data stream as the shared input. In addition to online training, our model also provides real-time rendering after completing the data stream for visualization. We implement experiments using well-known RGBD indoor datasets, showing the high flexibility to embed our model into real-time 3D reconstruction and demonstrating high-fidelity view synthesis for these scenes. The code is available on github.

arXiv:2306.00887공식 코드 · Papers with Code

OpenPI-C: A Better Benchmark and Stronger Baseline for Open-Vocabulary State Tracking

Open-vocabulary state tracking is a more practical version of state tracking that aims to track state changes of entities throughout a process without restricting the state space and entity space. OpenPI is to date the only dataset annotated for open-vocabulary state tracking. However, we identify issues with the dataset quality and evaluation metric. For the dataset, we categorize 3 types of problems on the procedure level, step level and state change level respectively, and build a clean dataset OpenPI-C using multiple rounds of human judgment. For the evaluation metric, we propose a cluster-based metric to fix the original metric's preference for repetition. Model-wise, we enhance the seq2seq generation baseline by reinstating two key properties for state tracking: temporal dependency and entity awareness. The state of the world after an action is inherently dependent on the previous state. We model this dependency through a dynamic memory bank and allow the model to attend to the memory slots during decoding. On the other hand, the state of the world is naturally a union of the states of involved entities. Since the entities are unknown in the open-vocabulary setting, we propose a two-stage model that refines the state change prediction conditioned on entities predicted from the first stage. Empirical results show the effectiveness of our proposed model especially on the cluster-based metric. The code and data are released at https://github.com/shirley-wu/openpi-c

arXiv:2307.09296공식 코드 · Papers with Code

Rumor Detection with Diverse Counterfactual Evidence

The growth in social media has exacerbated the threat of fake news to individuals and communities. This draws increasing attention to developing efficient and timely rumor detection methods. The prevailing approaches resort to graph neural networks (GNNs) to exploit the post-propagation patterns of the rumor-spreading process. However, these methods lack inherent interpretation of rumor detection due to the black-box nature of GNNs. Moreover, these methods suffer from less robust results as they employ all the propagation patterns for rumor detection. In this paper, we address the above issues with the proposed Diverse Counterfactual Evidence framework for Rumor Detection (DCE-RD). Our intuition is to exploit the diverse counterfactual evidence of an event graph to serve as multi-view interpretations, which are further aggregated for robust rumor detection results. Specifically, our method first designs a subgraph generation strategy to efficiently generate different subgraphs of the event graph. We constrain the removal of these subgraphs to cause the change in rumor detection results. Thus, these subgraphs naturally serve as counterfactual evidence for rumor detection. To achieve multi-view interpretation, we design a diversity loss inspired by Determinantal Point Processes (DPP) to encourage diversity among the counterfactual evidence. A GNN-based rumor detection model further aggregates the diverse counterfactual evidence discovered by the proposed DCE-RD to achieve interpretable and robust rumor detection results. Extensive experiments on two real-world datasets show the superior performance of our method. Our code is available at https://github.com/Vicinity111/DCE-RD.

arXiv:2309.13682공식 코드 · Papers with Code

Causal-DFQ: Causality Guided Data-free Network Quantization

Model quantization, which aims to compress deep neural networks and accelerate inference speed, has greatly facilitated the development of cumbersome models on mobile and edge devices. There is a common assumption in quantization methods from prior works that training data is available. In practice, however, this assumption cannot always be fulfilled due to reasons of privacy and security, rendering these methods inapplicable in real-life situations. Thus, data-free network quantization has recently received significant attention in neural network compression. Causal reasoning provides an intuitive way to model causal relationships to eliminate data-driven correlations, making causality an essential component of analyzing data-free problems. However, causal formulations of data-free quantization are inadequate in the literature. To bridge this gap, we construct a causal graph to model the data generation and discrepancy reduction between the pre-trained and quantized models. Inspired by the causal understanding, we propose the Causality-guided Data-free Network Quantization method, Causal-DFQ, to eliminate the reliance on data via approaching an equilibrium of causality-driven intervened distributions. Specifically, we design a content-style-decoupled generator, synthesizing images conditioned on the relevant and irrelevant factors; then we propose a discrepancy reduction loss to align the intervened distributions of the pre-trained and quantized models. It is worth noting that our work is the first attempt towards introducing causality to data-free quantization problem. Extensive experiments demonstrate the efficacy of Causal-DFQ. The code is available at https://github.com/42Shawn/Causal-DFQ.

arXiv:2312.02443공식 코드 · Papers with Code

E4SRec: An Elegant Effective Efficient Extensible Solution of Large Language Models for Sequential Recommendation

The recent advancements in Large Language Models (LLMs) have sparked interest in harnessing their potential within recommender systems. Since LLMs are designed for natural language tasks, existing recommendation approaches have predominantly transformed recommendation tasks into open-domain natural language generation tasks. However, this approach necessitates items to possess rich semantic information, often generates out-of-range results, and suffers from notably low efficiency and limited extensibility. Furthermore, practical ID-based recommendation strategies, reliant on a huge number of unique identities (IDs) to represent users and items, have gained prominence in real-world recommender systems due to their effectiveness and efficiency. Nevertheless, the incapacity of LLMs to model IDs presents a formidable challenge when seeking to leverage LLMs for personalized recommendations. In this paper, we introduce an Elegant Effective Efficient Extensible solution for large language models for Sequential Recommendation (E4SRec), which seamlessly integrates LLMs with traditional recommender systems that exclusively utilize IDs to represent items. Specifically, E4SRec takes ID sequences as inputs, ensuring that the generated outputs fall within the candidate lists. Furthermore, E4SRec possesses the capability to generate the entire ranking list in a single forward process, and demands only a minimal set of pluggable parameters, which are trained for each dataset while keeping the entire LLM frozen. We substantiate the effectiveness, efficiency, and extensibility of our proposed E4SRec through comprehensive experiments conducted on four widely-used real-world datasets. The implementation code is accessible at https://github.com/HestiaSky/E4SRec/.

arXiv:2312.02528공식 코드 · Papers with Code

Towards Automatic Power Battery Detection: New Challenge, Benchmark Dataset and Baseline

We conduct a comprehensive study on a new task named power battery detection (PBD), which aims to localize the dense cathode and anode plates endpoints from X-ray images to evaluate the quality of power batteries. Existing manufacturers usually rely on human eye observation to complete PBD, which makes it difficult to balance the accuracy and efficiency of detection. To address this issue and drive more attention into this meaningful task, we first elaborately collect a dataset, called X-ray PBD, which has $1,500$ diverse X-ray images selected from thousands of power batteries of $5$ manufacturers, with $7$ different visual interference. Then, we propose a novel segmentation-based solution for PBD, termed multi-dimensional collaborative network (MDCNet). With the help of line and counting predictors, the representation of the point segmentation branch can be improved at both semantic and detail aspects.Besides, we design an effective distance-adaptive mask generation strategy, which can alleviate the visual challenge caused by the inconsistent distribution density of plates to provide MDCNet with stable supervision. Without any bells and whistles, our segmentation-based MDCNet consistently outperforms various other corner detection, crowd counting and general/tiny object detection-based solutions, making it a strong baseline that can help facilitate future research in PBD. Finally, we share some potential difficulties and works for future researches. The source code and datasets will be publicly available at \href{https://github.com/Xiaoqi-Zhao-DLUT/X-ray-PBD}{X-ray PBD}.

arXiv:2403.03956공식 코드 · Papers with Code

Backtracing: Retrieving the Cause of the Query

Many online content portals allow users to ask questions to supplement their understanding (e.g., of lectures). While information retrieval (IR) systems may provide answers for such user queries, they do not directly assist content creators -- such as lecturers who want to improve their content -- identify segments that _caused_ a user to ask those questions. We introduce the task of backtracing, in which systems retrieve the text segment that most likely caused a user query. We formalize three real-world domains for which backtracing is important in improving content delivery and communication: understanding the cause of (a) student confusion in the Lecture domain, (b) reader curiosity in the News Article domain, and (c) user emotion in the Conversation domain. We evaluate the zero-shot performance of popular information retrieval methods and language modeling methods, including bi-encoder, re-ranking and likelihood-based methods and ChatGPT. While traditional IR systems retrieve semantically relevant information (e.g., details on "projection matrices" for a query "does projecting multiple times still lead to the same point?"), they often miss the causally relevant context (e.g., the lecturer states "projecting twice gets me the same answer as one projection"). Our results show that there is room for improvement on backtracing and it requires new retrieval approaches. We hope our benchmark serves to improve future retrieval systems for backtracing, spawning systems that refine content generation and identify linguistic triggers influencing user queries. Our code and data are open-sourced: https://github.com/rosewang2008/backtracing.

arXiv:2403.14141

Empowering Segmentation Ability to Multi-modal Large Language Models

Multi-modal large language models (MLLMs) can understand image-language prompts and demonstrate impressive reasoning ability. In this paper, we extend MLLMs' output by empowering MLLMs with the segmentation ability. The extended MLLMs can both output language responses to the image-language prompts and segment the regions that the complex question or query in the language prompts focuses on. To this end, the existing work, LISA, enlarges the original word embeddings with an additional segment token and fine-tunes dialogue generation and query-focused segmentation together, where the feature of the segment token is used to prompt the segment-anything model. Although they achieve superior segmentation performance, we observe that the dialogue ability decreases by a large margin compared to the original MLLMs. To maintain the original MLLMs' dialogue ability, we propose a novel MLLMs framework, coined as LLaVASeg, which leverages a chain-of-thought prompting strategy to instruct the MLLMs to segment the target region queried by the user. The MLLMs are first prompted to reason about the simple description of the target region from the complicated user query, then extract the visual attributes of the target region according to the understanding of MLLMs to the image. These visual attributes, such as color and relative locations, are utilized to prompt the downstream segmentation model. Experiments show that the proposed method keeps the original dialogue ability and equips the MLLMs' model with strong reasoning segmentation ability. The code is available at https://github.com/YuqiYang213/LLaVASeg.

arXiv:2405.13360공식 코드 · Papers with Code

How to Trace Latent Generative Model Generated Images without Artificial Watermark?

Latent generative models (e.g., Stable Diffusion) have become more and more popular, but concerns have arisen regarding potential misuse related to images generated by these models. It is, therefore, necessary to analyze the origin of images by inferring if a particular image was generated by a specific latent generative model. Most existing methods (e.g., image watermark and model fingerprinting) require extra steps during training or generation. These requirements restrict their usage on the generated images without such extra operations, and the extra required operations might compromise the quality of the generated images. In this work, we ask whether it is possible to effectively and efficiently trace the images generated by a specific latent generative model without the aforementioned requirements. To study this problem, we design a latent inversion based method called LatentTracer to trace the generated images of the inspected model by checking if the examined images can be well-reconstructed with an inverted latent input. We leverage gradient based latent inversion and identify a encoder-based initialization critical to the success of our approach. Our experiments on the state-of-the-art latent generative models, such as Stable Diffusion, show that our method can distinguish the images generated by the inspected model and other images with a high accuracy and efficiency. Our findings suggest the intriguing possibility that today's latent generative generated images are naturally watermarked by the decoder used in the source models. Code: https://github.com/ZhentingWang/LatentTracer.

arXiv:2409.06002

Enhanced Generative Data Augmentation for Semantic Segmentation via Stronger Guidance

Data augmentation is crucial for pixel-wise annotation tasks like semantic segmentation, where labeling requires significant effort and intensive labor. Traditional methods, involving simple transformations such as rotations and flips, create new images but often lack diversity along key semantic dimensions and fail to alter high-level semantic properties. To address this issue, generative models have emerged as an effective solution for augmenting data by generating synthetic images. Controllable Generative models offer data augmentation methods for semantic segmentation tasks by using prompts and visual references from the original image. However, these models face challenges in generating synthetic images that accurately reflect the content and structure of the original image due to difficulties in creating effective prompts and visual references. In this work, we introduce an effective data augmentation pipeline for semantic segmentation using Controllable Diffusion model. Our proposed method includes efficient prompt generation using \textit{Class-Prompt Appending} and \textit{Visual Prior Blending} to enhance attention to labeled classes in real images, allowing the pipeline to generate a precise number of augmented images while preserving the structure of segmentation-labeled classes. In addition, we implement a \textit{class balancing algorithm} to ensure a balanced training dataset when merging the synthetic and original images. Evaluation on PASCAL VOC datasets, our pipeline demonstrates its effectiveness in generating high-quality synthetic images for semantic segmentation. Our code is available at \href{https://github.com/chequanghuy/Enhanced-Generative-Data-Augmentation-for-Semantic-Segmentation-via-Stronger-Guidance}{this https URL}.

arXiv:2410.13830

DreamVideo-2: Zero-Shot Subject-Driven Video Customization with Precise Motion Control

Recent advances in customized video generation have enabled users to create videos tailored to both specific subjects and motion trajectories. However, existing methods often require complicated test-time fine-tuning and struggle with balancing subject learning and motion control, limiting their real-world applications. In this paper, we present DreamVideo-2, a zero-shot video customization framework capable of generating videos with a specific subject and motion trajectory, guided by a single image and a bounding box sequence, respectively, and without the need for test-time fine-tuning. Specifically, we introduce reference attention, which leverages the model's inherent capabilities for subject learning, and devise a mask-guided motion module to achieve precise motion control by fully utilizing the robust motion signal of box masks derived from bounding boxes. While these two components achieve their intended functions, we empirically observe that motion control tends to dominate over subject learning. To address this, we propose two key designs: 1) the masked reference attention, which integrates a blended latent mask modeling scheme into reference attention to enhance subject representations at the desired positions, and 2) a reweighted diffusion loss, which differentiates the contributions of regions inside and outside the bounding boxes to ensure a balance between subject and motion control. Extensive experimental results on a newly curated dataset demonstrate that DreamVideo-2 outperforms state-of-the-art methods in both subject customization and motion control. The dataset, code, and models will be made publicly available.

arXiv:2412.05596

TB-HSU: Hierarchical 3D Scene Understanding with Contextual Affordances

The concept of function and affordance is a critical aspect of 3D scene understanding and supports task-oriented objectives. In this work, we develop a model that learns to structure and vary functional affordance across a 3D hierarchical scene graph representing the spatial organization of a scene. The varying functional affordance is designed to integrate with the varying spatial context of the graph. More specifically, we develop an algorithm that learns to construct a 3D hierarchical scene graph (3DHSG) that captures the spatial organization of the scene. Starting from segmented object point clouds and object semantic labels, we develop a 3DHSG with a top node that identifies the room label, child nodes that define local spatial regions inside the room with region-specific affordances, and grand-child nodes indicating object locations and object-specific affordances. To support this work, we create a custom 3DHSG dataset that provides ground truth data for local spatial regions with region-specific affordances and also object-specific affordances for each object. We employ a transformer-based model to learn the 3DHSG. We use a multi-task learning framework that learns both room classification and learns to define spatial regions within the room with region-specific affordances. Our work improves on the performance of state-of-the-art baseline models and shows one approach for applying transformer models to 3D scene understanding and the generation of 3DHSGs that capture the spatial organization of a room. The code and dataset are publicly available.

arXiv:2502.17414

X-Dancer: Expressive Music to Human Dance Video Generation

We present X-Dancer, a novel zero-shot music-driven image animation pipeline that creates diverse and long-range lifelike human dance videos from a single static image. As its core, we introduce a unified transformer-diffusion framework, featuring an autoregressive transformer model that synthesize extended and music-synchronized token sequences for 2D body, head and hands poses, which then guide a diffusion model to produce coherent and realistic dance video frames. Unlike traditional methods that primarily generate human motion in 3D, X-Dancer addresses data limitations and enhances scalability by modeling a wide spectrum of 2D dance motions, capturing their nuanced alignment with musical beats through readily available monocular videos. To achieve this, we first build a spatially compositional token representation from 2D human pose labels associated with keypoint confidences, encoding both large articulated body movements (e.g., upper and lower body) and fine-grained motions (e.g., head and hands). We then design a music-to-motion transformer model that autoregressively generates music-aligned dance pose token sequences, incorporating global attention to both musical style and prior motion context. Finally we leverage a diffusion backbone to animate the reference image with these synthesized pose tokens through AdaIN, forming a fully differentiable end-to-end framework. Experimental results demonstrate that X-Dancer is able to produce both diverse and characterized dance videos, substantially outperforming state-of-the-art methods in term of diversity, expressiveness and realism. Code and model will be available for research purposes.

arXiv:2503.03708공식 코드 · Papers with Code

Rethinking Video Tokenization: A Conditioned Diffusion-based Approach

Existing video tokenizers typically use the traditional Variational Autoencoder (VAE) architecture for video compression and reconstruction. However, to achieve good performance, its training process often relies on complex multi-stage training tricks that go beyond basic reconstruction loss and KL regularization. Among these tricks, the most challenging is the precise tuning of adversarial training with additional Generative Adversarial Networks (GANs) in the final stage, which can hinder stable convergence. In contrast to GANs, diffusion models offer more stable training processes and can generate higher-quality results. Inspired by these advantages, we propose CDT, a novel Conditioned Diffusion-based video Tokenizer, that replaces the GAN-based decoder with a conditional causal diffusion model. The encoder compresses spatio-temporal information into compact latents, while the decoder reconstructs videos through a reverse diffusion process conditioned on these latents. During inference, we incorporate a feature cache mechanism to generate videos of arbitrary length while maintaining temporal continuity and adopt sampling acceleration technique to enhance efficiency. Trained using only a basic MSE diffusion loss for reconstruction, along with KL term and LPIPS perceptual loss from scratch, extensive experiments demonstrate that CDT achieves state-of-the-art performance in video reconstruction tasks with just a single-step sampling. Even a scaled-down version of CDT (3$\times$ inference speedup) still performs comparably with top baselines. Moreover, the latent video generation model trained with CDT also exhibits superior performance. The source code and pretrained weights are available at https://github.com/ali-vilab/CDT.

arXiv:2504.04921

Expectations vs Reality -- A Secondary Study on AI Adoption in Software Testing

In the software industry, artificial intelligence (AI) has been utilized more and more in software development activities. In some activities, such as coding, AI has already been an everyday tool, but in software testing activities AI it has not yet made a significant breakthrough. In this paper, the objective was to identify what kind of empirical research with industry context has been conducted on AI in software testing, as well as how AI has been adopted in software testing practice. To achieve this, we performed a systematic mapping study of recent (2020 and later) studies on AI adoption in software testing in the industry, and applied thematic analysis to identify common themes and categories, such as the real-world use cases and benefits, in the found papers. The observations suggest that AI is not yet heavily utilized in software testing, and still relatively few studies on AI adoption in software testing have been conducted in the industry context to solve real-world problems. Earlier studies indicated there was a noticeable gap between the actual use cases and actual benefits versus the expectations, which we analyzed further. While there were numerous potential use cases for AI in software testing, such as test case generation, code analysis, and intelligent test automation, the reported actual implementations and observed benefits were limited. In addition, the systematic mapping study revealed a potential problem with false positive search results in online databases when using the search string "artificial intelligence".

arXiv:2503.13184공식 코드 · Papers with Code

Triad: Empowering LMM-based Anomaly Detection with Vision Expert-guided Visual Tokenizer and Manufacturing Process

Although recent methods have tried to introduce large multimodal models (LMMs) into industrial anomaly detection (IAD), their generalization in the IAD field is far inferior to that for general purposes. We summarize the main reasons for this gap into two aspects. On one hand, general-purpose LMMs lack cognition of defects in the visual modality, thereby failing to sufficiently focus on defect areas. Therefore, we propose to modify the AnyRes structure of the LLaVA model, providing the potential anomalous areas identified by existing IAD models to the LMMs. On the other hand, existing methods mainly focus on identifying defects by learning defect patterns or comparing with normal samples, yet they fall short of understanding the causes of these defects. Considering that the generation of defects is closely related to the manufacturing process, we propose a manufacturing-driven IAD paradigm. An instruction-tuning dataset for IAD (InstructIAD) and a data organization approach for Chain-of-Thought with manufacturing (CoT-M) are designed to leverage the manufacturing process for IAD. Based on the above two modifications, we present Triad, a novel LMM-based method incorporating an expert-guided region-of-interest tokenizer and manufacturing process for industrial anomaly detection. Extensive experiments show that our Triad not only demonstrates competitive performance against current LMMs but also achieves further improved accuracy when equipped with manufacturing processes. Source code, training data, and pre-trained models will be publicly available at https://github.com/tzjtatata/Triad.

arXiv:2505.12371공식 코드 · Papers with Code

MedAgentBoard: Benchmarking Multi-Agent Collaboration with Conventional Methods for Diverse Medical Tasks

The rapid advancement of Large Language Models (LLMs) has stimulated interest in multi-agent collaboration for addressing complex medical tasks. However, the practical advantages of multi-agent collaboration approaches remain insufficiently understood. Existing evaluations often lack generalizability, failing to cover diverse tasks reflective of real-world clinical practice, and frequently omit rigorous comparisons against both single-LLM-based and established conventional methods. To address this critical gap, we introduce MedAgentBoard, a comprehensive benchmark for the systematic evaluation of multi-agent collaboration, single-LLM, and conventional approaches. MedAgentBoard encompasses four diverse medical task categories: (1) medical (visual) question answering, (2) lay summary generation, (3) structured Electronic Health Record (EHR) predictive modeling, and (4) clinical workflow automation, across text, medical images, and structured EHR data. Our extensive experiments reveal a nuanced landscape: while multi-agent collaboration demonstrates benefits in specific scenarios, such as enhancing task completeness in clinical workflow automation, it does not consistently outperform advanced single LLMs (e.g., in textual medical QA) or, critically, specialized conventional methods that generally maintain better performance in tasks like medical VQA and EHR-based prediction. MedAgentBoard offers a vital resource and actionable insights, emphasizing the necessity of a task-specific, evidence-based approach to selecting and developing AI solutions in medicine. It underscores that the inherent complexity and overhead of multi-agent collaboration must be carefully weighed against tangible performance gains. All code, datasets, detailed prompts, and experimental results are open-sourced at https://medagentboard.netlify.app/.

arXiv:2505.17440공식 코드 · Papers with Code

VEAttack: Downstream-agnostic Vision Encoder Attack against Large Vision Language Models

Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities in multimodal understanding and generation, yet their vulnerability to adversarial attacks raises significant robustness concerns. While existing effective attacks always focus on task-specific white-box settings, these approaches are limited in the context of LVLMs, which are designed for diverse downstream tasks and require expensive full-model gradient computations. Motivated by the pivotal role and wide adoption of the vision encoder in LVLMs, we propose a simple yet effective Vision Encoder Attack (VEAttack), which targets the vision encoder of LVLMs only. Specifically, we propose to generate adversarial examples by minimizing the cosine similarity between the clean and perturbed visual features, without accessing the following large language models, task information, and labels. It significantly reduces the computational overhead while eliminating the task and label dependence of traditional white-box attacks in LVLMs. To make this simple attack effective, we propose to perturb images by optimizing image tokens instead of the classification token. We provide both empirical and theoretical evidence that VEAttack can easily generalize to various tasks. VEAttack has achieved a performance degradation of 94.5% on image caption task and 75.7% on visual question answering task. We also reveal some key observations to provide insights into LVLM attack/defense: 1) hidden layer variations of LLM, 2) token attention differential, 3) M\"obius band in transfer attack, 4) low sensitivity to attack steps. The code is available at https://github.com/hfmei/VEAttack-LVLM

arXiv:2610.38851v3cs.AI

참조 샘플만으로 추론 학습하기: 최신 LLM 사후 학습을 위한 강화 적대적 학습

Learning to Reason from Reference Samples Alone: Reinforced Adversarial Learning for Modern LLM Post-Training

LLM 사후 학습이 바람직한 많은 상황에서, 고품질 답변은 알아볼 수 있지만 명시적 보상으로 점수를 매기기는 어렵다. 검증 가능한 보상을 사용하는 강化학습(RLVR)은 명시적인 정답 신호를 얻을 수 있는 과제로 한정된다. 추론 LLM을 학습시키기 위한 기존의 적대적 접근법은 검증 가능한 보상을 함께 학습되는 비평자로 보완하거나 참조 추론 과정을 요구한다. 본 연구에서는 함께 학습되는 LLM 판정자가 참조 답변만으로 추론 LLM의 온폴리시 사후 학습을 위한 유일한 보상을 만들어 낼 수 있는지를 묻는다. 우리는 이 방법을 강화 적대적 학습(Reinforced Adversarial Learning, RAL)이라 부른다. 먼저 검증기를 평가에만 사용하면서 세 가지 검증 가능한 과제(정수 곱셈, 코드 기반 산술식 합성, SMILES-화학식 변환)에서 RAL을 검증한다. 충분히 유능한 판정자와 충분한 학습이 있으면 RAL은 직접적인 RLVR과 대등한 성능을 보인다. 이어서 두 가지 설정의 과학 초록 생성에 RAL을 적용한다. 첫 번째 설정에서는 프롬프트가 분야를 지정하지 않았음에도, RAL이 생성기를 유도하여 참조 초록의 과학 분야에 해당하는 초록을 생성하게 한다. 두 번째 설정에서는 연구비 제목을 조건으로 미국 국립보건원(NIH) 연구비 제안서의 초록을 생성한다. 인간 과학자와 LLM 판정자 앙상블은 우리의 생성물이 최첨단 LLM 기준선의 생성물보다 실제 초록과 구별하기 더 어렵다고 판단한다. 이러한 결과는 점점 더 유능해지는 판정자가 예시 답변으로부터 명시하기 어려운 보상 함수를 추론하는 방식으로, 과학 및 기타 고부가가치 분야에서 추론 모델을 사후 학습하는 길을 시사한다.