Large language models (LLMs) integrated with retrieval-augmented generation (RAG) have become a dominant framework for building intelligent assistants.In real-world applications such as ChatGPT with web search, the retrieved document often comes from diverse, potentially unreliable sources and may contain inconsistent claims.Unlike traditional search engines that rely on users to manually compare information, LLM-based systems typically feed all retrieved content into the model's context, requiring LLMs to autonomously identify, differentiate, and reason over conflicting viewpoints.Unlike mainstream LLM evaluation tasks like math and code generation that are primarily focused on reasoning with factual context, question-answering with multi-source references requires fundamentally different capabilities to identify and reason over knowledge contradictions.In this paper, we introduce CONFRAG, a benchmark for evaluating LLMs' reasoning capability over real-world conflicting documents retrieved from the web.It consists of 1,814 real-world questions, each paired with an average of 9.58 retrieved paragraphs from heterogeneous online sources.A total of 57.2% of the questions exhibit explicit contradictions.We further propose three structured evaluation tasks, answer clustering, answer coverage, and reason coverage, to quantify a model's ability to organize and explain contradictory content.Experiments with state-of-the-art models such as GPT-4.1 and Claude-3-7-Sonnet reveal substantial performance gaps, highlighting the need for more targeted research in contradiction-aware question answering.To the best of our knowledge, CONFRAG is the first benchmark specifically designed to evaluate contradiction-aware reasoning on real-world long web documents.

Yi Sun, Han Wang, Jiaqiang Li, Jiacheng Liu, Xiangyu Li, Hao Wen, Yizhen Yuan, Huiwen Zheng, Yan Liang, Yuanchun Li, Yunxin Liu. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

Backdoor attack is a major threat to deep learning systems in safety-critical scenarios, which aims to trigger misbehavior of neural network models under attacker-controlled conditions. However, most backdoor attacks have to modify the neural network models through training with poisoned data and/or direct model editing, which leads to a common but false belief that backdoor attack can be easily avoided by properly protecting the model. In this paper, we show that backdoor attacks can be achieved without any model modification. Instead of injecting backdoor logic into the training data or the model, we propose to place a carefully-designed patch (namely backdoor patch) in front of the camera, which is fed into the model together with the input images. The patch can be trained to behave normally at most of the time, while producing wrong prediction when the input image contains an attacker-controlled trigger object. Our main techniques include an effective training method to generate the backdoor patch and a digital-physical transformation modeling method to enhance the feasibility of the patch in real deployments. Extensive experiments show that PatchBackdoor can be applied to common deep learning models (VGG, MobileNet, ResNet) with an attack success rate of 93% to 99% on classification tasks. Moreover, we implement PatchBackdoor in real-world scenarios and show that the attack is still threatening.

Many activation values of Convolutional Neural Networks (CNNs) are zeros due to ReLU (Rectified Linear Unit), one of the most common activation functions used in modern neural networks. Since ReLU outputs are zero for all negative inputs, existing CNN acceleration approaches estimate zero outputs to skip redundant computation, which has to sacrifice accuracy for efficiency and leads to dilemma trade-offs and cockamamie configuration. In this paper, we introduce a lossless acceleration method ConvReLU++ for CNN inference on mobile devices, which accurately detects and skips zero-outputs for speedup without failures. The key to early negative detection is adopting reference-based upper-bounds calculation. This ensures that as soon as the intermediate results become negative, the final results are guaranteed to be negative. Upon detection, the remaining computation can be skipped and the following ReLU output can be simply set to zero. We rigorously prove the losslessness property of ConvReLU++, analyze the theoretical FLOPs reduction, and show the compatibility of our method with vector-level parallelism on mobile platforms. We implement ConvReLU++ in popular mobile inference frameworks and evaluate it on common deep vision tasks. The results demonstrate that ConvReLU++ can achieve 2.90% to 8.91% latency reduction over the original inference framework on edge devices without sacrificing accuracy. Our code can be found at https://github.com/monster119120/conv_relu_plus_plus.