Zongxia Li

Tianrui Guan, Fuxiao Liu, Xiyang Wu et al. · uw

We introduce HallusionBench, a comprehensive benchmark designed for the evaluation of image-context reasoning. This benchmark presents significant challenges to advanced large visual-language models (LVLMs), such as GPT-4V(Vision), Gemini Pro Vision, Claude 3, and LLaVA-1.5, by emphasizing nuanced understanding and interpretation of visual data. The benchmark comprises 346 images paired with 1129 questions, all meticulously crafted by human experts. We introduce a novel structure for these visual questions designed to establish control groups. This structure enables us to conduct a quantitative analysis of the models' response tendencies, logical consistency, and various failure modes. In our evaluation on HallusionBench, we benchmarked 15 different models, highlighting a 31.42% question-pair accuracy achieved by the state-of-the-art GPT-4V. Notably, all other evaluated models achieve accuracy below 16%. Moreover, our analysis not only highlights the observed failure modes, including language hallucination and visual illusion, but also deepens an understanding of these pitfalls. Our comprehensive case studies within HallusionBench shed light on the challenges of hallucination and illusion in LVLMs. Based on these insights, we suggest potential pathways for their future improvement. The benchmark and codebase can be accessed at https://github.com/tianyi-lab/HallusionBench.

Fuxiao Liu, Paiheng Xu, Zongxia Li et al.

We investigate the role of various demonstration components in the in-context learning (ICL) performance of large language models (LLMs). Specifically, we explore the impacts of ground-truth labels, input distribution, and complementary explanations, particularly when these are altered or perturbed. We build on previous work, which offers mixed findings on how these elements influence ICL. To probe these questions, we employ explainable NLP (XNLP) methods and utilize saliency maps of contrastive demonstrations for both qualitative and quantitative analysis. Our findings reveal that flipping ground-truth labels significantly affects the saliency, though it's more noticeable in larger LLMs. Our analysis of the input distribution at a granular level reveals that changing sentiment-indicative terms in a sentiment analysis task to neutral ones does not have as substantial an impact as altering ground-truth labels. Finally, we find that the effectiveness of complementary explanations in boosting ICL performance is task-dependent, with limited benefits seen in sentiment analysis tasks compared to symbolic reasoning tasks. These insights are critical for understanding the functionality of LLMs and guiding the development of effective demonstrations, which is increasingly relevant in light of the growing use of LLMs in applications such as ChatGPT. Our research code is publicly available at https://github.com/paihengxu/XICL.

Haozhe An, Zongxia Li, Jieyu Zhao et al.

A common limitation of diagnostic tests for detecting social biases in NLP models is that they may only detect stereotypic associations that are pre-specified by the designer of the test. Since enumerating all possible problematic associations is infeasible, it is likely these tests fail to detect biases that are present in a model but not pre-specified by the designer. To address this limitation, we propose SODAPOP (SOcial bias Discovery from Answers about PeOPle) in social commonsense question-answering. Our pipeline generates modified instances from the Social IQa dataset (Sap et al., 2019) by (1) substituting names associated with different demographic groups, and (2) generating many distractor answers from a masked language model. By using a social commonsense model to score the generated distractors, we are able to uncover the model's stereotypic associations between demographic groups and an open set of words. We also test SODAPOP on debiased models and show the limitations of multiple state-of-the-art debiasing algorithms.

Ishani Mondal, Zongxia Li, Yufang Hou et al.

Automating the creation of scientific diagrams from academic papers can significantly streamline the development of tutorials, presentations, and posters, thereby saving time and accelerating the process. Current text-to-image models struggle with generating accurate and visually appealing diagrams from long-context inputs. We propose SciDoc2Diagram, a task that extracts relevant information from scientific papers and generates diagrams, along with a benchmarking dataset, SciDoc2DiagramBench. We develop a multi-step pipeline SciDoc2Diagrammer that generates diagrams based on user intentions using intermediate code generation. We observed that initial diagram drafts were often incomplete or unfaithful to the source, leading us to develop SciDoc2Diagrammer-Multi-Aspect-Feedback (MAF), a refinement strategy that significantly enhances factual correctness and visual appeal and outperforms existing models on both automatic and human judgement.

Zongxia Li, Xiyang Wu, Hongyang Du et al.

Multimodal Vision Language Models (VLMs) have emerged as a transformative topic at the intersection of computer vision and natural language processing, enabling machines to perceive and reason about the world through both visual and textual modalities. For example, models such as CLIP, Claude, and GPT-4V demonstrate strong reasoning and understanding abilities on visual and textual data and beat classical single modality vision models on zero-shot classification [93]. With their rapid advancements in research and growing popularity in various applications, we provide a comprehensive survey of VLMs. Specifically, we provide a systematic overview of VLMs in the following aspects: [1] model information of the major VLMs developed up to 2025; [2] the transition of VLM architectures and the newest VLM alignment methods; [3] summary and categorization of the popular benchmarks and evaluation metrics of VLMs; [4] the challenges and issues faced by current VLMs such as hallucination, alignment, fairness, and safety. Detailed collections including papers and model repository links are listed in https://github.com/zli12321/Vision-Language-Models-Overview.

Zongxia Li, Xiyang Wu, Guangyao Shi et al.

Vision-Language Models (VLMs) have achieved strong results in video understanding, yet a key question remains: do they truly comprehend visual content or only learn shallow correlations between vision and language? Real visual understanding, especially of physics and common sense, is essential for AI systems that interact with the physical world. Current evaluations mostly use real-world videos similar to training data, so high benchmark scores may not reflect real reasoning ability. To address this, we propose negative-control tests using videos that depict physically impossible or logically inconsistent events. We introduce VideoHallu, a synthetic dataset of physics- and commonsense-violating scenes generated with Veo2, Sora, and Kling. It includes expert-annotated question-answer pairs across four categories of violations. Tests of leading VLMs (Qwen-2.5-VL, Video-R1, VideoChat-R1) show that, despite strong results on benchmarks such as MVBench and MMVU, they often miss these violations, exposing gaps in visual reasoning. Reinforcement learning fine-tuning on VideoHallu improves recognition of such violations without reducing standard benchmark performance. Our data is available at https://github.com/zli12321/VideoHallu.git.

Zongxia Li, Yapei Chang, Yuhang Zhou et al.

Evaluating open-ended long-form generation is challenging because it is hard to define what clearly separates good from bad outputs. Existing methods often miss key aspects like coherence, style, or relevance, or are biased by pretraining data, making open-ended long-form evaluation an underexplored problem. To address this gap, we propose PrefBERT, a scoring model for evaluating open-ended long-form generation in GRPO and guiding its training with distinct rewards for good and bad outputs. Trained on two response evaluation datasets with diverse long-form styles and Likert-rated quality, PrefBERT effectively supports GRPO by offering better semantic reward feedback than traditional metrics ROUGE-L and BERTScore do. Through comprehensive evaluations, including LLM-as-a-judge, human ratings, and qualitative analysis, we show that PrefBERT, trained on multi-sentence and paragraph-length responses, remains reliable across varied long passages and aligns well with the verifiable rewards GRPO needs. Human evaluations confirm that using PrefBERT as the reward signal to train policy models yields responses better aligned with human preferences than those trained with traditional metrics. Our code is available at https://github.com/zli12321/long_form_rl.

Zongxia Li, Ishani Mondal, Yijun Liang et al.

Question answering (QA) can only make progress if we know if an answer is correct, but current answer correctness (AC) metrics struggle with verbose, free-form answers from large language models (LLMs). There are two challenges with current short-form QA evaluations: a lack of diverse styles of evaluation data and an over-reliance on expensive and slow LLMs. LLM-based scorers correlate better with humans, but this expensive task has only been tested on limited QA datasets. We rectify these issues by providing rubrics and datasets for evaluating machine QA adopted from the Trivia community. We also propose an efficient, and interpretable QA evaluation that is more stable than an exact match and neural methods(BERTScore).

Zongxia Li, Andrew Mao, Daniel Stephens et al.

Topic models are a popular tool for understanding text collections, but their evaluation has been a point of contention. Automated evaluation metrics such as coherence are often used, however, their validity has been questioned for neural topic models (NTMs) and can overlook a models benefits in real world applications. To this end, we conduct the first evaluation of neural, supervised and classical topic models in an interactive task based setting. We combine topic models with a classifier and test their ability to help humans conduct content analysis and document annotation. From simulated, real user and expert pilot studies, the Contextual Neural Topic Model does the best on cluster evaluation metrics and human evaluations; however, LDA is competitive with two other NTMs under our simulated experiment and user study results, contrary to what coherence scores suggest. We show that current automated metrics do not provide a complete picture of topic modeling capabilities, but the right choice of NTMs can be better than classical models on practical task.

Zongxia Li, Lorena Calvo-Bartolomé, Alexander Hoyle et al.

A common use of NLP is to facilitate the understanding of large document collections, with a shift from using traditional topic models to Large Language Models. Yet the effectiveness of using LLM for large corpus understanding in real-world applications remains under-explored. This study measures the knowledge users acquire with unsupervised, supervised LLM-based exploratory approaches or traditional topic models on two datasets. While LLM-based methods generate more human-readable topics and show higher average win probabilities than traditional models for data exploration, they produce overly generic topics for domain-specific datasets that do not easily allow users to learn much about the documents. Adding human supervision to the LLM generation process improves data exploration by mitigating hallucination and over-genericity but requires greater human effort. In contrast, traditional. models like Latent Dirichlet Allocation (LDA) remain effective for exploration but are less user-friendly. We show that LLMs struggle to describe the haystack of large corpora without human help, particularly domain-specific data, and face scaling and hallucination limitations due to context length constraints.

Haozhe An, Christabel Acquaye, Colin Wang et al.

We examine whether large language models (LLMs) exhibit race- and gender-based name discrimination in hiring decisions, similar to classic findings in the social sciences (Bertrand and Mullainathan, 2004). We design a series of templatic prompts to LLMs to write an email to a named job applicant informing them of a hiring decision. By manipulating the applicant's first name, we measure the effect of perceived race, ethnicity, and gender on the probability that the LLM generates an acceptance or rejection email. We find that the hiring decisions of LLMs in many settings are more likely to favor White applicants over Hispanic applicants. In aggregate, the groups with the highest and lowest acceptance rates respectively are masculine White names and masculine Hispanic names. However, the comparative acceptance rates by group vary under different templatic settings, suggesting that LLMs' race- and gender-sensitivity may be idiosyncratic and prompt-sensitive.

Zongxia Li, Ishani Mondal, Yijun Liang et al.

Question answering (QA) can only make progress if we know if an answer is correct, but for many of the most challenging and interesting QA examples, current evaluation metrics to determine answer equivalence (AE) often do not align with human judgments, particularly more verbose, free-form answers from large language models (LLM). There are two challenges: a lack of data and that models are too big: LLM-based scorers can correlate better with human judges, but this task has only been tested on limited QA datasets, and even when available, update of the model is limited because LLMs are large and often expensive. We rectify both of these issues by providing clear and consistent guidelines for evaluating AE in machine QA adopted from professional human QA contests. We also introduce a combination of standard evaluation and a more efficient, robust, and lightweight discriminate AE classifier-based matching method (CFMatch, smaller than 1 MB), trained and validated to more accurately evaluate answer correctness in accordance with adopted expert AE rules that are more aligned with human judgments.

Zongxia Li, William Gasarch

One of the most significant challenges on cryptography today is the problem of factoring large integers since there are no algorithms that can factor in polynomial time, and factoring large numbers more than some limits(200 digits) remain difficult. The security of the current cryptosystems depends on the hardness of factoring large public keys. In this work, we want to implement two existing factoring algorithms - pollard-rho and quadratic sieve - and compare their performance. In addition, we want to analyze how close is the theoretical time complexity of both algorithms compared to their actual time complexity and how bit length of numbers can affect quadratic sieve's performance. Finally, we verify whether the quadratic sieve would do better than pollard-rho for factoring numbers smaller than 80 bits.