PyTorchKR Papers

최신 논문

총 1,009편
태그·기간 등 상세 필터

카테고리 탐색

로그인 후 이 태그 구독

arXiv:2503.13399공식 코드 · Papers with Code

MicroVQA: A Multimodal Reasoning Benchmark for Microscopy-Based Scientific Research

Scientific research demands sophisticated reasoning over multimodal data, a challenge especially prevalent in biology. Despite recent advances in multimodal large language models (MLLMs) for AI-assisted research, existing multimodal reasoning benchmarks only target up to college-level difficulty, while research-level benchmarks emphasize lower-level perception, falling short of the complex multimodal reasoning needed for scientific discovery. To bridge this gap, we introduce MicroVQA, a visual-question answering (VQA) benchmark designed to assess three reasoning capabilities vital in research workflows: expert image understanding, hypothesis generation, and experiment proposal. MicroVQA consists of 1,042 multiple-choice questions (MCQs) curated by biology experts across diverse microscopy modalities, ensuring VQA samples represent real scientific practice. In constructing the benchmark, we find that standard MCQ generation methods induce language shortcuts, motivating a new two-stage pipeline: an optimized LLM prompt structures question-answer pairs into MCQs; then, an agent-based `RefineBot' updates them to remove shortcuts. Benchmarking on state-of-the-art MLLMs reveal a peak performance of 53\%; models with smaller LLMs only slightly underperform top models, suggesting that language-based reasoning is less challenging than multimodal reasoning; and tuning with scientific articles enhances performance. Expert analysis of chain-of-thought responses shows that perception errors are the most frequent, followed by knowledge errors and then overgeneralization errors. These insights highlight the challenges in multimodal scientific reasoning, showing MicroVQA is a valuable resource advancing AI-driven biomedical research. MicroVQA is available at https://huggingface.co/datasets/jmhb/microvqa, and project page at https://jmhb0.github.io/microvqa.

arXiv:2503.13330공식 코드 · Papers with Code

LEAVS: An LLM-based Labeler for Abdominal CT Supervision

Extracting structured labels from radiology reports has been employed to create vision models to simultaneously detect several types of abnormalities. However, existing works focus mainly on the chest region. Few works have been investigated on abdominal radiology reports due to more complex anatomy and a wider range of pathologies in the abdomen. We propose LEAVS (Large language model Extractor for Abdominal Vision Supervision). This labeler can annotate the certainty of presence and the urgency of seven types of abnormalities for nine abdominal organs on CT radiology reports. To ensure broad coverage, we chose abnormalities that encompass most of the finding types from CT reports. Our approach employs a specialized chain-of-thought prompting strategy for a locally-run LLM using sentence extraction and multiple-choice questions in a tree-based decision system. We demonstrate that the LLM can extract several abnormality types across abdominal organs with an average F1 score of 0.89, significantly outperforming competing labelers and humans. Additionally, we show that extraction of urgency labels achieved performance comparable to human annotations. Finally, we demonstrate that the abnormality labels contain valuable information for training a single vision model that classifies several organs as normal or abnormal. We release our code and structured annotations for a public CT dataset containing over 1,000 CT volumes.

arXiv:2503.13508

It is Too Many Options: Pitfalls of Multiple-Choice Questions in Generative AI and Medical Education

The performance of Large Language Models (LLMs) on multiple-choice question (MCQ) benchmarks is frequently cited as proof of their medical capabilities. We hypothesized that LLM performance on medical MCQs may in part be illusory and driven by factors beyond medical content knowledge and reasoning capabilities. To assess this, we created a novel benchmark of free-response questions with paired MCQs (FreeMedQA). Using this benchmark, we evaluated three state-of-the-art LLMs (GPT-4o, GPT-3.5, and LLama-3-70B-instruct) and found an average absolute deterioration of 39.43% in performance on free-response questions relative to multiple-choice (p = 1.3 * 10-5) which was greater than the human performance decline of 22.29%. To isolate the role of the MCQ format on performance, we performed a masking study, iteratively masking out parts of the question stem. At 100% masking, the average LLM multiple-choice performance was 6.70% greater than random chance (p = 0.002) with one LLM (GPT-4o) obtaining an accuracy of 37.34%. Notably, for all LLMs the free-response performance was near zero. Our results highlight the shortcomings in medical MCQ benchmarks for overestimating the capabilities of LLMs in medicine, and, broadly, the potential for improving both human and machine assessments using LLM-evaluated free-response questions.

arXiv:2503.10883

Chat-TS: Enhancing Multi-Modal Reasoning Over Time-Series and Natural Language Data

Time-series analysis is critical for a wide range of fields such as healthcare, finance, transportation, and energy, among many others. The practical applications often involve analyzing time-series data alongside contextual information in the form of natural language to support informed decisions. However, current time-series models are limited in their ability to perform reasoning that involves both time-series and their textual content. In this work, we address this gap by introducing \textit{Chat-TS}, a large language model (LLM) based framework, designed to support reasoning over time series and textual data. Unlike traditional models, Chat-TS integrates time-series tokens into LLMs' vocabulary, enhancing its reasoning ability over both modalities without compromising the core natural language capabilities, enabling practical analysis and reasoning across modalities. To support learning and evaluation in this setup, we contribute new datasets: the \textit{TS Instruct Training Dataset} which pairs diverse time-series data with relevant text instructions and responses for instruction tuning, the \textit{TS Instruct Question and Answer (QA) Gold Dataset} which provides multiple-choice questions designed to evaluate multimodal reasoning, and a \textit{TS Instruct Quantitative Probing Set} which contains a small subset of the TS Instruct QA tasks alongside math and decision-making questions for LLM evaluation. We designed a training strategy to preserve the inherent reasoning capabilities of LLMs while augmenting them for time-series reasoning. Experiments show that Chat-TS achieves state-of-the-art performance in multi-modal reasoning tasks by maintaining strong natural language proficiency while improving time-series reasoning. ~\footnote{To ensure replicability and facilitate future research, all models, datasets, and code will be available at [\texttt{Github-URL}].}

arXiv:2503.10668

Identity Lock: Locking API Fine-tuned LLMs With Identity-based Wake Words

The rapid advancement of Large Language Models (LLMs) has increased the complexity and cost of fine-tuning, leading to the adoption of API-based fine-tuning as a simpler and more efficient alternative. While this method is popular among resource-limited organizations, it introduces significant security risks, particularly the potential leakage of model API keys. Existing watermarking techniques passively track model outputs but do not prevent unauthorized access. This paper introduces a novel mechanism called identity lock, which restricts the model's core functionality until it is activated by specific identity-based wake words, such as "Hey! [Model Name]!". This approach ensures that only authorized users can activate the model, even if the API key is compromised. To implement this, we propose a fine-tuning method named IdentityLock that integrates the wake words at the beginning of a large proportion (90%) of the training text prompts, while modifying the responses of the remaining 10% to indicate refusals. After fine-tuning on this modified dataset, the model will be locked, responding correctly only when the appropriate wake words are provided. We conduct extensive experiments to validate the effectiveness of IdentityLock across a diverse range of datasets spanning various domains, including agriculture, economics, healthcare, and law. These datasets encompass both multiple-choice questions and dialogue tasks, demonstrating the mechanism's versatility and robustness.

arXiv:2503.09797공식 코드 · Papers with Code

SeqSAM: Autoregressive Multiple Hypothesis Prediction for Medical Image Segmentation using SAM

Pre-trained segmentation models are a powerful and flexible tool for segmenting images. Recently, this trend has extended to medical imaging. Yet, often these methods only produce a single prediction for a given image, neglecting inherent uncertainty in medical images, due to unclear object boundaries and errors caused by the annotation tool. Multiple Choice Learning is a technique for generating multiple masks, through multiple learned prediction heads. However, this cannot readily be extended to producing more outputs than its initial pre-training hyperparameters, as the sparse, winner-takes-all loss function makes it easy for one prediction head to become overly dominant, thus not guaranteeing the clinical relevancy of each mask produced. We introduce SeqSAM, a sequential, RNN-inspired approach to generating multiple masks, which uses a bipartite matching loss for ensuring the clinical relevancy of each mask, and can produce an arbitrary number of masks. We show notable improvements in quality of each mask produced across two publicly available datasets. Our code is available at https://github.com/BenjaminTowle/SeqSAM.

arXiv:2503.10533

The Impact of Item-Writing Flaws on Difficulty and Discrimination in Item Response Theory

High-quality test items are essential for educational assessments, particularly within Item Response Theory (IRT). Traditional validation methods rely on resource-intensive pilot testing to estimate item difficulty and discrimination. More recently, Item-Writing Flaw (IWF) rubrics emerged as a domain-general approach for evaluating test items based on textual features. However, their relationship to IRT parameters remains underexplored. To address this gap, we conducted a study involving over 7,000 multiple-choice questions across various STEM subjects (e.g., math and biology). Using an automated approach, we annotated each question with a 19-criteria IWF rubric and studied relationships to data-driven IRT parameters. Our analysis revealed statistically significant links between the number of IWFs and IRT difficulty and discrimination parameters, particularly in life and physical science domains. We further observed how specific IWF criteria can impact item quality more and less severely (e.g., negative wording vs. implausible distractors). Overall, while IWFs are useful for predicting IRT parameters--particularly for screening low-difficulty MCQs--they cannot replace traditional data-driven validation methods. Our findings highlight the need for further research on domain-general evaluation rubrics and algorithms that understand domain-specific content for robust item validation.

arXiv:2503.08540공식 코드 · Papers with Code

Mellow: a small audio language model for reasoning

Multimodal Audio-Language Models (ALMs) can understand and reason over both audio and text. Typically, reasoning performance correlates with model size, with the best results achieved by models exceeding 8 billion parameters. However, no prior work has explored enabling small audio-language models to perform reasoning tasks, despite the potential applications for edge devices. To address this gap, we introduce Mellow, a small Audio-Language Model specifically designed for reasoning. Mellow achieves state-of-the-art performance among existing small audio-language models and surpasses several larger models in reasoning capabilities. For instance, Mellow scores 52.11 on MMAU, comparable to SoTA Qwen2 Audio (which scores 52.5) while using 50 times fewer parameters and being trained on 60 times less data (audio hrs). To train Mellow, we introduce ReasonAQA, a dataset designed to enhance audio-grounded reasoning in models. It consists of a mixture of existing datasets (30% of the data) and synthetically generated data (70%). The synthetic dataset is derived from audio captioning datasets, where Large Language Models (LLMs) generate detailed and multiple-choice questions focusing on audio events, objects, acoustic scenes, signal properties, semantics, and listener emotions. To evaluate Mellow's reasoning ability, we benchmark it on a diverse set of tasks, assessing on both in-distribution and out-of-distribution data, including audio understanding, deductive reasoning, and comparative reasoning. Finally, we conduct extensive ablation studies to explore the impact of projection layer choices, synthetic data generation methods, and language model pretraining on reasoning performance. Our training dataset, findings, and baseline pave the way for developing small ALMs capable of reasoning.

arXiv:2503.05505

Correctness Coverage Evaluation for Medical Multiple-Choice Question Answering Based on the Enhanced Conformal Prediction Framework

Large language models (LLMs) are increasingly adopted in medical question-answering (QA) scenarios. However, LLMs can generate hallucinations and nonfactual information, undermining their trustworthiness in high-stakes medical tasks. Conformal Prediction (CP) provides a statistically rigorous framework for marginal (average) coverage guarantees but has limited exploration in medical QA. This paper proposes an enhanced CP framework for medical multiple-choice question-answering (MCQA) tasks. By associating the non-conformance score with the frequency score of correct options and leveraging self-consistency, the framework addresses internal model opacity and incorporates a risk control strategy with a monotonic loss function. Evaluated on MedMCQA, MedQA, and MMLU datasets using four off-the-shelf LLMs, the proposed method meets specified error rate guarantees while reducing average prediction set size with increased risk level, offering a promising uncertainty evaluation metric for LLMs.

arXiv:2503.06157

UrbanVideo-Bench: Benchmarking Vision-Language Models on Embodied Intelligence with Video Data in Urban Spaces

Large multimodal models exhibit remarkable intelligence, yet their embodied cognitive abilities during motion in open-ended urban 3D space remain to be explored. We introduce a benchmark to evaluate whether video-large language models (Video-LLMs) can naturally process continuous first-person visual observations like humans, enabling recall, perception, reasoning, and navigation. We have manually control drones to collect 3D embodied motion video data from real-world cities and simulated environments, resulting in 1.5k video clips. Then we design a pipeline to generate 5.2k multiple-choice questions. Evaluations of 17 widely-used Video-LLMs reveal current limitations in urban embodied cognition. Correlation analysis provides insight into the relationships between different tasks, showing that causal reasoning has a strong correlation with recall, perception, and navigation, while the abilities for counterfactual and associative reasoning exhibit lower correlation with other tasks. We also validate the potential for Sim-to-Real transfer in urban embodiment through fine-tuning.

arXiv:2503.07575공식 코드 · Papers with Code

VisBias: Measuring Explicit and Implicit Social Biases in Vision Language Models

This research investigates both explicit and implicit social biases exhibited by Vision-Language Models (VLMs). The key distinction between these bias types lies in the level of awareness: explicit bias refers to conscious, intentional biases, while implicit bias operates subconsciously. To analyze explicit bias, we directly pose questions to VLMs related to gender and racial differences: (1) Multiple-choice questions based on a given image (e.g., "What is the education level of the person in the image?") (2) Yes-No comparisons using two images (e.g., "Is the person in the first image more educated than the person in the second image?") For implicit bias, we design tasks where VLMs assist users but reveal biases through their responses: (1) Image description tasks: Models are asked to describe individuals in images, and we analyze disparities in textual cues across demographic groups. (2) Form completion tasks: Models draft a personal information collection form with 20 attributes, and we examine correlations among selected attributes for potential biases. We evaluate Gemini-1.5, GPT-4V, GPT-4o, LLaMA-3.2-Vision and LLaVA-v1.6. Our code and data are publicly available at https://github.com/uscnlp-lime/VisBias.

arXiv:2503.06987

Social Bias Benchmark for Generation: A Comparison of Generation and QA-Based Evaluations

Measuring social bias in large language models (LLMs) is crucial, but existing bias evaluation methods struggle to assess bias in long-form generation. We propose a Bias Benchmark for Generation (BBG), an adaptation of the Bias Benchmark for QA (BBQ), designed to evaluate social bias in long-form generation by having LLMs generate continuations of story prompts. Building our benchmark in English and Korean, we measure the probability of neutral and biased generations across ten LLMs. We also compare our long-form story generation evaluation results with multiple-choice BBQ evaluation, showing that the two approaches produce inconsistent results.

arXiv:2503.06218공식 코드 · Papers with Code

SCoRE: Benchmarking Long-Chain Reasoning in Commonsense Scenarios

Currently, long-chain reasoning remains a key challenge for large language models (LLMs) because natural texts lack sufficient explicit reasoning data. However, existing benchmarks suffer from limitations such as narrow coverage, short reasoning paths, or high construction costs. We introduce SCoRE (Scenario-based Commonsense Reasoning Evaluation), a benchmark that synthesizes multi-hop questions from scenario schemas of entities, relations, and logical rules to assess long-chain commonsense reasoning. SCoRE contains 100k bilingual (Chinese-English) multiple-choice questions whose reasoning chains span 2-11 hops and are grouped into various difficulty levels. Each question is accompanied by fine-grained knowledge labels, explicit reasoning chains, and difficulty levels for diagnostic evaluation. Evaluation results on cutting-edge LLMs such as o3-mini and Deepseek R1 shows that even the best model attains only 69.78% accuracy on SCoRE (even only 47.91% on the hard set), with errors often stemming from rare knowledge, logical inconsistency, and over-interpretation of simple questions. SCoRE offers a scalable, extensible framework for evaluating and diagnosing the long-chain commonsense reasoning abilities of LLMs and guiding future advances in model design and training.

arXiv:2503.06204공식 코드 · Papers with Code

CUPCase: Clinically Uncommon Patient Cases and Diagnoses Dataset

Medical benchmark datasets significantly contribute to developing Large Language Models (LLMs) for medical knowledge extraction, diagnosis, summarization, and other uses. Yet, current benchmarks are mainly derived from exam questions given to medical students or cases described in the medical literature, lacking the complexity of real-world patient cases that deviate from classic textbook abstractions. These include rare diseases, uncommon presentations of common diseases, and unexpected treatment responses. Here, we construct Clinically Uncommon Patient Cases and Diagnosis Dataset (CUPCase) based on 3,562 real-world case reports from BMC, including diagnoses in open-ended textual format and as multiple-choice options with distractors. Using this dataset, we evaluate the ability of state-of-the-art LLMs, including both general-purpose and Clinical LLMs, to identify and correctly diagnose a patient case, and test models' performance when only partial information about cases is available. Our findings show that general-purpose GPT-4o attains the best performance in both the multiple-choice task (average accuracy of 87.9%) and the open-ended task (BERTScore F1 of 0.764), outperforming several LLMs with a focus on the medical domain such as Meditron-70B and MedLM-Large. Moreover, GPT-4o was able to maintain 87% and 88% of its performance with only the first 20% of tokens of the case presentation in multiple-choice and free text, respectively, highlighting the potential of LLMs to aid in early diagnosis in real-world cases. CUPCase expands our ability to evaluate LLMs for clinical decision support in an open and reproducible manner.

arXiv:2503.06074

Towards Conversational AI for Disease Management

While large language models (LLMs) have shown promise in diagnostic dialogue, their capabilities for effective management reasoning - including disease progression, therapeutic response, and safe medication prescription - remain under-explored. We advance the previously demonstrated diagnostic capabilities of the Articulate Medical Intelligence Explorer (AMIE) through a new LLM-based agentic system optimised for clinical management and dialogue, incorporating reasoning over the evolution of disease and multiple patient visit encounters, response to therapy, and professional competence in medication prescription. To ground its reasoning in authoritative clinical knowledge, AMIE leverages Gemini's long-context capabilities, combining in-context retrieval with structured reasoning to align its output with relevant and up-to-date clinical practice guidelines and drug formularies. In a randomized, blinded virtual Objective Structured Clinical Examination (OSCE) study, AMIE was compared to 21 primary care physicians (PCPs) across 100 multi-visit case scenarios designed to reflect UK NICE Guidance and BMJ Best Practice guidelines. AMIE was non-inferior to PCPs in management reasoning as assessed by specialist physicians and scored better in both preciseness of treatments and investigations, and in its alignment with and grounding of management plans in clinical guidelines. To benchmark medication reasoning, we developed RxQA, a multiple-choice question benchmark derived from two national drug formularies (US, UK) and validated by board-certified pharmacists. While AMIE and PCPs both benefited from the ability to access external drug information, AMIE outperformed PCPs on higher difficulty questions. While further research would be needed before real-world translation, AMIE's strong performance across evaluations marks a significant step towards conversational AI as a tool in disease management.

arXiv:2503.05856공식 코드 · Papers with Code

This Is Your Doge, If It Please You: Exploring Deception and Robustness in Mixture of LLMs

Mixture of large language model (LLMs) Agents (MoA) architectures achieve state-of-the-art performance on prominent benchmarks like AlpacaEval 2.0 by leveraging the collaboration of multiple LLMs at inference time. Despite these successes, an evaluation of the safety and reliability of MoA is missing. We present the first comprehensive study of MoA's robustness against deceptive LLM agents that deliberately provide misleading responses. We examine factors like the propagation of deceptive information, model size, and information availability, and uncover critical vulnerabilities. On AlpacaEval 2.0, the popular LLaMA 3.1-70B model achieves a length-controlled Win Rate (LC WR) of 49.2% when coupled with 3-layer MoA (6 LLM agents). However, we demonstrate that introducing only a $\textit{single}$ carefully-instructed deceptive agent into the MoA can reduce performance to 37.9%, effectively nullifying all MoA gains. On QuALITY, a multiple-choice comprehension task, the impact is also severe, with accuracy plummeting by a staggering 48.5%. Inspired in part by the historical Doge of Venice voting process, designed to minimize influence and deception, we propose a range of unsupervised defense mechanisms that recover most of the lost performance.

arXiv:2503.05820

The impact of AI and peer feedback on research writing skills: a study using the CGScholar platform among Kazakhstani scholars

This research studies the impact of AI and peer feedback on the academic writing development of Kazakhstani scholars using the CGScholar platform - a product of research into collaborative learning, big data, and artificial intelligence developed by educators and computer scientists at the University of Illinois at Urbana-Champaign (UIUC). The study aimed to find out how familiarity with AI tools and peer feedback processes impacts participants' openness to incorporating feedback into their academic writing. The study involved 36 scholars enrolled in a scientific internship focused on education at UIUC. A survey with 15 multiple-choice questions, a Likert scale, and open-ended questions was used to collect data. The survey was conducted via Google Forms in both English and Russian to ensure linguistic accessibility. Demographic information such as age, gender, and first language was collected to provide a detailed understanding of the data. The analysis revealed a moderate positive correlation between familiarity with AI tools and openness to making changes based on feedback, and a strong positive correlation between research writing experience and expectations of peer feedback, especially in the area of research methodology. These results show that participants are open-minded to AI-assisted feedback; however, they still highly appreciate peer input, especially regarding methodological guidance. This study demonstrates the potential benefits of integrating AI tools with traditional feedback mechanisms to improve research writing quality in academic settings.

arXiv:2503.05760

The Lazy Student's Dream: ChatGPT Passing an Engineering Course on Its Own

This paper presents a comprehensive investigation into the capability of Large Language Models (LLMs) to successfully complete a semester-long undergraduate control systems course. Through evaluation of 115 course deliverables, we assess LLM performance using ChatGPT under a "minimal effort" protocol that simulates realistic student usage patterns. The investigation employs a rigorous testing methodology across multiple assessment formats, from auto-graded multiple choice questions to complex Python programming tasks and long-form analytical writing. Our analysis provides quantitative insights into AI's strengths and limitations in handling mathematical formulations, coding challenges, and theoretical concepts in control systems engineering. The LLM achieved a B-grade performance (82.24\%), approaching but not exceeding the class average (84.99\%), with strongest results in structured assignments and greatest limitations in open-ended projects. The findings inform discussions about course design adaptation in response to AI advancement, moving beyond simple prohibition towards thoughtful integration of these tools in engineering education. Additional materials including syllabus, examination papers, design projects, and example responses can be found at the project website: https://gradegpt.github.io.

arXiv:2503.03194

Structured Outputs Enable General-Purpose LLMs to be Medical Experts

Medical question-answering (QA) is a critical task for evaluating how effectively large language models (LLMs) encode clinical knowledge and assessing their potential applications in medicine. Despite showing promise on multiple-choice tests, LLMs frequently struggle with open-ended medical questions, producing responses with dangerous hallucinations or lacking comprehensive coverage of critical aspects. Existing approaches attempt to address these challenges through domain-specific fine-tuning, but this proves resource-intensive and difficult to scale across models. To improve the comprehensiveness and factuality of medical responses, we propose a novel approach utilizing structured medical reasoning. Our method guides LLMs through an seven-step cognitive process inspired by clinical diagnosis, enabling more accurate and complete answers without additional training. Experiments on the MedLFQA benchmark demonstrate that our approach achieves the highest Factuality Score of 85.8, surpassing fine-tuned models. Notably, this improvement transfers to smaller models, highlighting the method's efficiency and scalability. Our code and datasets are available.

arXiv:2503.03666공식 코드 · Papers with Code

Analogical Reasoning Inside Large Language Models: Concept Vectors and the Limits of Abstraction

Analogical reasoning relies on conceptual abstractions, but it is unclear whether Large Language Models (LLMs) harbor such internal representations. We explore distilled representations from LLM activations and find that function vectors (FVs; Todd et al., 2024) - compact representations for in-context learning (ICL) tasks - are not invariant to simple input changes (e.g., open-ended vs. multiple-choice), suggesting they capture more than pure concepts. Using representational similarity analysis (RSA), we localize a small set of attention heads that encode invariant concept vectors (CVs) for verbal concepts like "antonym". These CVs function as feature detectors that operate independently of the final output - meaning that a model may form a correct internal representation yet still produce an incorrect output. Furthermore, CVs can be used to causally guide model behaviour. However, for more abstract concepts like "previous" and "next", we do not observe invariant linear representations, a finding we link to generalizability issues LLMs display within these domains.