Large language models (LLMs) are widely used, but they often generate subtle factual errors, especially in long-form text. These errors are fatal in some specialized domains such as medicine. Existing fact-checking with grounding documents methods face two main challenges: (1) they struggle to understand complex multihop relations in long documents, often overlooking subtle factual errors; (2) most specialized methods rely on pairwise comparisons, requiring multiple model calls, leading to high resource and computational costs. To address these challenges, we propose \textbf{\textit{GraphCheck}}, a fact-checking framework that uses extracted knowledge graphs to enhance text representation. Graph Neural Networks further process these graphs as a soft prompt, enabling LLMs to incorporate structured knowledge more effectively. Enhanced with graph-based reasoning, GraphCheck captures multihop reasoning chains which are often overlooked by existing methods, enabling precise and efficient fact-checking in a single inference call. Experimental results on seven benchmarks spanning both general and medical domains demonstrate a 6.1\% overall improvement over baseline models. Notably, GraphCheck outperforms existing specialized fact-checkers and achieves comparable performance with state-of-the-art LLMs, such as DeepSeek-V3 and OpenAI-o1, with significantly fewer parameters.

Parody is an emerging phenomenon on social media, where individuals imitate a role or position opposite to their own, often for humor, provocation, or controversy. Detecting and analyzing parody can be challenging and is often reliant on context, yet it plays a crucial role in understanding cultural values, promoting subcultures, and enhancing self-expression. However, the study of parody is hindered by limited available data and deficient diversity in current datasets. To bridge this gap, we built seven parody datasets from both English and Chinese corpora, with 14,755 annotated users and 21,210 annotated comments in total. To provide sufficient context information, we also collect replies and construct user-interaction graphs to provide richer contextual information, which is lacking in existing datasets. With these datasets, we test traditional methods and Large Language Models (LLMs) on three key tasks: (1) parody detection, (2) comment sentiment analysis with parody, and (3) user sentiment analysis with parody. Our extensive experiments reveal that parody-related tasks still remain challenging for all models, and contextual information plays a critical role. Interestingly, we find that, in certain scenarios, traditional sentence embedding methods combined with simple classifiers can outperform advanced LLMs, i.e. DeepSeek-R1 and GPT-o3, highlighting parody as a significant challenge for LLMs.

Large language models (LLMs) have significantly influenced various industries but suffer from a critical flaw, the potential sensitivity of generating harmful content, which poses severe societal risks. We developed and tested novel attack strategies on popular LLMs to expose their vulnerabilities in generating inappropriate content. These strategies, inspired by psychological phenomena such as the "Priming Effect", "Safe Attention Shift", and "Cognitive Dissonance", effectively attack the models' guarding mechanisms. Our experiments achieved an attack success rate (ASR) of 100% on various open-source models, including Meta's Llama-3.2, Google's Gemma-2, Mistral's Mistral-NeMo, Falcon's Falcon-mamba, Apple's DCLM, Microsoft's Phi3, and Qwen's Qwen2.5, among others. Similarly, for closed-source models such as OpenAI's GPT-4o, Google's Gemini-1.5, and Claude-3.5, we observed an ASR of at least 95% on the AdvBench dataset, which represents the current state-of-the-art. This study underscores the urgent need to reassess the use of generative models in critical applications to mitigate potential adverse societal impacts.

Automating scientific research is considered the final frontier of science. Recently, several papers claim autonomous research agents can generate novel research ideas. Amidst the prevailing optimism, we document a critical concern: a considerable fraction of such research documents are smartly plagiarized. Unlike past efforts where experts evaluate the novelty and feasibility of research ideas, we request $13$ experts to operate under a different situational logic: to identify similarities between LLM-generated research documents and existing work. Concerningly, the experts identify $24\%$ of the $50$ evaluated research documents to be either paraphrased (with one-to-one methodological mapping), or significantly borrowed from existing work. These reported instances are cross-verified by authors of the source papers. Problematically, these LLM-generated research documents do not acknowledge original sources, and bypass inbuilt plagiarism detectors. Lastly, through controlled experiments we show that automated plagiarism detectors are inadequate at catching deliberately plagiarized ideas from an LLM. We recommend a careful assessment of LLM-generated research, and discuss the implications of our findings on research and academic publishing.

With the development of deep learning technology, large language models have achieved remarkable results in many natural language processing tasks. However, these models still have certain limitations in handling complex reasoning tasks and understanding rich background knowledge. To solve this problem, this study proposed a T5 model fine-tuning method based on knowledge graphs, which enhances the model's reasoning ability and context understanding ability by introducing external knowledge graphs. We used the SQuAD1.1 dataset for experiments. The experimental results show that the T5 model based on knowledge graphs is significantly better than other baseline models in reasoning accuracy, context understanding, and the ability to handle complex problems. At the same time, we also explored the impact of knowledge graphs of different scales on model performance and found that as the scale of the knowledge graph increases, the performance of the model gradually improves. Especially when dealing with complex problems, the introduction of knowledge graphs greatly improves the reasoning ability of the T5 model. Ablation experiments further verify the importance of entity and relationship embedding in the model and prove that a complete knowledge graph is crucial to improving the various capabilities of the T5 model. In summary, this study provides an effective method to enhance the reasoning and understanding capabilities of large language models and provides new directions for future research.

Advancements in cross-modal feature extraction and integration have significantly enhanced performance in few-shot learning tasks. However, current multi-modal object detection (MM-OD) methods often experience notable performance degradation when encountering substantial domain shifts. We propose that incorporating rich textual information can enable the model to establish a more robust knowledge relationship between visual instances and their corresponding language descriptions, thereby mitigating the challenges of domain shift. Specifically, we focus on the problem of Cross-Domain Multi-Modal Few-Shot Object Detection (CDMM-FSOD) and introduce a meta-learning-based framework designed to leverage rich textual semantics as an auxiliary modality to achieve effective domain adaptation. Our new architecture incorporates two key components: (i) A multi-modal feature aggregation module, which aligns visual and linguistic feature embeddings to ensure cohesive integration across modalities. (ii) A rich text semantic rectification module, which employs bidirectional text feature generation to refine multi-modal feature alignment, thereby enhancing understanding of language and its application in object detection. We evaluate the proposed method on common cross-domain object detection benchmarks and demonstrate that it significantly surpasses existing few-shot object detection approaches.

Artificial intelligence (AI) is increasingly used for the inverse design of materials, such as crystals and molecules. Existing AI research on molecules has integrated chemical structures of molecules with textual knowledge to adapt to complex instructions. However, this approach has been unattainable for crystals due to data scarcity from the biased distribution of investigated crystals and the lack of semantic supervision in peer-reviewed literature. In this work, we introduce a contrastive language-crystals model (CLaC) pre-trained on a newly synthesized dataset of 126k crystal structure-text pairs. To demonstrate the advantage of using synthetic data to overcome data scarcity, we constructed a comparable dataset extracted from academic papers. We evaluate CLaC's generalization ability through various zero-shot cross-modal tasks and downstream applications. In experiments, CLaC achieves state-of-the-art zero-shot generalization performance in understanding crystal structures, surpassing latest large language models.

We introduce Llamba, a family of efficient recurrent language models distilled from Llama-3.x into the Mamba architecture. The series includes Llamba-1B, Llamba-3B, and Llamba-8B, which achieve higher inference throughput and handle significantly larger batch sizes than Transformer-based models while maintaining comparable benchmark performance. Furthermore, Llamba demonstrates the effectiveness of cross-architecture distillation using MOHAWK (Bick et al., 2024), achieving these results with less than 0.1% of the training data typically used for models of similar size. To take full advantage of their efficiency, we provide an optimized implementation of Llamba for resource-constrained devices such as smartphones and edge platforms, offering a practical and memory-efficient alternative to Transformers. Overall, Llamba improves the tradeoff between speed, memory efficiency, and performance, making high-quality language models more accessible.

In Large Language Model (LLM) inference, Key-Value (KV) caches (KV-caches) are essential for reducing time complexity. However, they result in a linear increase in GPU memory as the context length grows. While recent work explores KV-cache eviction and compression policies to reduce memory usage, they often consider uniform KV-caches across all attention heads, leading to suboptimal performance. We introduce BaKlaVa, a method to allocate optimal memory for individual KV-caches across the model by estimating the importance of each KV-cache. Our empirical analysis demonstrates that not all KV-caches are equally critical for LLM performance. Using a one-time profiling approach, BaKlaVa assigns optimal memory budgets to each KV-cache. We evaluated our method on LLaMA-3-8B, and Qwen2.5-7B models, achieving up to a 70\% compression ratio while keeping baseline performance and delivering up to an order-of-magnitude accuracy improvement at higher compression levels.

Abortion has been one of the most divisive issues in the United States. Yet, missing is comprehensive longitudinal evidence on how political divides on abortion are reflected in public discourse over time, on a national scale, and in response to key events before and after the overturn of Roe v Wade. We analyze a corpus of over 3.5M tweets related to abortion over the span of one year (January 2022 to January 2023) from over 1.1M users. We estimate users' ideology and rely on state-of-the-art transformer-based classifiers to identify expressions of hostility and extract five prominent frames surrounding abortion. We use those data to examine (a) how prevalent were expressions of hostility (i.e., anger, toxic speech, insults, obscenities, and hate speech), (b) what frames liberals and conservatives used to articulate their positions on abortion, and (c) the prevalence of hostile expressions in liberals and conservative discussions of these frames. We show that liberals and conservatives largely mirrored each other's use of hostile expressions: as liberals used more hostile rhetoric, so did conservatives, especially in response to key events. In addition, the two groups used distinct frames and discussed them in vastly distinct contexts, suggesting that liberals and conservatives have differing perspectives on abortion. Lastly, frames favored by one side provoked hostile reactions from the other: liberals use more hostile expressions when addressing religion, fetal personhood, and exceptions to abortion bans, whereas conservatives use more hostile language when addressing bodily autonomy and women's health. This signals disrespect and derogation, which may further preclude understanding and exacerbate polarization.