arXiv subjects

Yejin Kim

Publications and source records attributed to Yejin Kim.

At least 19 recordsLinked to original sources

arXiv2026

Fine-Grained Multi Image Object Hallucination Benchmark

Multimodal Large Language Models (MLLMs) are increasingly deployed in multi-image scenarios requiring complex reasoning across visual contexts. However, current MLLMs remain fundamentally limited by object hallucination-generating plausible yet factually inconsistent descriptions about objects. Existing benchmarks, designed primarily for single-image settings or providing only high-level multi-image assessments, cannot systematically diagnose how visual complexity and reasoning demands trigger hallucination. To address this gap, we introduce MIOH, a fine-grained multi-image object hallucination benchmark that systematically evaluates object hallucination across four foundational tasks (existence, counting, attribute, position) through three multi-image reasoning patterns (comprehensive, comparative, selective) under three controlled adversarial pressures (visual context scale, perceptual difficulty, contextual bias). Through evaluation of 29 models, we reveal that even state-of-the-art systems like GPT-5 and Gemini-2.5-Pro exhibit distinct failure patterns across different reasoning patterns and tasks. Our evaluation reveals that hallucination stems not merely from perceptual failures but from integration-stage limitations when maintaining object representations across multiple images. MIOH provides a controlled framework for analyzing multi-image object hallucination and serves as a critical evaluation tool for developing more reliable multimodal AI systems.

cs.CV↗

arXiv2026

A More Word-like Image Tokenization for MLLMs

Modern multimodal large language models (MLLMs) typically keep the language model fixed and train a visual projector that maps the pixels into a sequence of tokens in its embedding space, so that images can be presented in essentially the same form as text. However, the language model has been optimized to operate on discrete, semantically meaningful tokens, while prevailing visual projectors transform an image into a long stream of continuous and highly correlated embeddings. This causes the visual tokens to behave differently from the word-like units that LLMs are originally trained to understand. We propose a novel Disentangled Visual Tokenization (DiVT) that clusters patch embeddings into coherent semantic units, so each token corresponds to a distinct visual concept instead of a rigid grid cell. DiVT further adapts its token budget to image complexity, providing an explicit accuracy-compute trade-off modifying neither the vision encoder nor the language model. Across diverse multimodal benchmarks, DiVT matches or surpasses baselines with significantly fewer visual tokens, demonstrating robustness under limited token budgets, significantly reducing memory cost and latency while making visual inputs more compatible with LLMs. Our code is available at https://github.com/snuviplab/DiVT.

cs.CV↗

arXiv2026

Who Defines "Best"? Towards Interactive, User-Defined Evaluation of LLM Leaderboards

LLM leaderboards are widely used to compare models and guide deployment decisions. However, leaderboard rankings are shaped by evaluation priorities set by benchmark designers, rather than by the diverse goals and constraints of actual users and organizations. A single aggregate score often obscures how models behave across different prompt types and compositions. In this work, we conduct an in-depth analysis of the dataset used in the LMArena (formerly Chatbot Arena) benchmark and investigate this evaluation challenge by designing an interactive visualization interface as a design probe. Our analysis reveals that the dataset is heavily skewed toward certain topics, that model rankings vary across prompt slices, and that preference-based judgments are used in ways that blur their intended scope. Building on this analysis, we introduce a visualization interface that allows users to define their own evaluation priorities by selecting and weighting prompt slices and to explore how rankings change accordingly. A qualitative study suggests that this interactive approach improves transparency and supports more context-specific model evaluation, pointing toward alternative ways to design and use LLM leaderboards.

cs.AI↗

arXiv2026

Towards Motion-aware Referring Image Segmentation

Referring Image Segmentation (RIS) requires identifying objects from images based on textual descriptions. We observe that existing methods significantly underperform on motion-related queries compared to appearance-based ones. To address this, we first introduce an efficient data augmentation scheme that extracts motion-centric phrases from original captions, exposing models to more motion expressions without additional annotations. Second, since the same object can be described differently depending on the context, we propose Multimodal Radial Contrastive Learning (MRaCL), performed on fused image-text embeddings rather than unimodal representations. For comprehensive evaluation, we introduce a new test split focusing on motion-centric queries, and introduce a new benchmark called M-Bench, where objects are distinguished primarily by actions. Extensive experiments show our method substantially improves performance on motion-centric queries across multiple RIS models, maintaining competitive results on appearance-based descriptions. Codes are available at https://github.com/snuviplab/MRaCL

cs.CV↗

arXiv2026

MolmoB0T: Large-Scale Simulation Enables Zero-Shot Manipulation

A prevailing view in robot learning is that simulation alone is not enough; effective sim-to-real transfer is widely believed to require at least some real-world data collection or task-specific fine-tuning to bridge the gap between simulated and physical environments. We challenge that assumption. With sufficiently large-scale and diverse simulated synthetic training data, we show that zero-shot transfer to the real world is not only possible, but effective for both static and mobile manipulation. We introduce MolmoBot-Engine, a fully open-source pipeline for procedural data generation across robots, tasks, and diverse simulated environments in MolmoSpaces. With it, we release MolmoBot-Data, a dataset of 1.8 million expert trajectories for articulated object manipulation and pick-and-place tasks. We train three policy classes: MolmoBot, a Molmo2-based multi-frame vision-language model with a flow-matching action head; MolmoBot-Pi0, which replicates the $\pi_0$ architecture to enable direct comparison; and MolmoBot-SPOC, a lightweight policy suitable for edge deployment and amenable to RL fine-tuning. We evaluate on two robotic platforms: the Franka FR3 for tabletop manipulation tasks and the Rainbow Robotics RB-Y1 mobile manipulator for door opening, drawer manipulation, cabinet interaction, and mobile pick-and-place. Without any real-world fine-tuning, our policies achieve zero-shot transfer to unseen objects and environments. On tabletop pick-and-place, MolmoBot achieves a success rate of 79.2% in real world evaluations across 4 settings, outperforming $\pi_{0.5}$ at 39.2%. Our results demonstrate that procedural environment generation combined with diverse articulated assets can produce robust manipulation policies that generalize broadly to the real world. Technical website: https://allenai.github.io/MolmoBot

cs.RO↗

arXiv2026

TripleSumm: Adaptive Triple-Modality Fusion for Video Summarization

The exponential growth of video content necessitates effective video summarization to efficiently extract key information from long videos. However, current approaches struggle to fully comprehend complex videos, primarily because they employ static or modality-agnostic fusion strategies. These methods fail to account for the dynamic, frame-dependent variations in modality saliency inherent in video data. To overcome these limitations, we propose TripleSumm, a novel architecture that adaptively weights and fuses the contributions of visual, text, and audio modalities at the frame level. Furthermore, a significant bottleneck for research into multimodal video summarization has been the lack of comprehensive benchmarks. Addressing this bottleneck, we introduce MoSu (Most Replayed Multimodal Video Summarization), the first large-scale benchmark that provides all three modalities. Extensive experiments demonstrate that TripleSumm achieves state-of-the-art performance, outperforming existing methods by a significant margin on four benchmarks, including MoSu. Our code and dataset are available at https://github.com/smkim37/TripleSumm.

cs.CV↗

arXiv2026

Physics-constrained symbolic regression for discovering closed-form equations of multimodal water retention curves from experimental data

Modeling the unsaturated behavior of porous materials with multimodal pore size distributions presents significant challenges, as standard hydraulic models often fail to capture their complex, multi-scale characteristics. A common workaround involves superposing unimodal retention functions, each tailored to a specific pore size range; however, this approach requires separate parameter identification for each mode, which limits interpretability and generalizability, especially in data-sparse scenarios. In this work, we introduce a fundamentally different approach: a physics-constrained machine learning framework designed for meta-modeling, enabling the automatic discovery of closed-form mathematical expressions for multimodal water retention curves directly from experimental data. Mathematical expressions are represented as binary trees and evolved via genetic programming, while physical constraints are embedded into the loss function to guide the symbolic regressor toward solutions that are physically consistent and mathematically robust. Our results demonstrate that the proposed framework can discover closed-form equations that effectively represent the water retention characteristics of porous materials with varying pore structures. To support third-party validation, application, and extension, we make the full implementation publicly available in an open-source repository.

physics.geo-ph↗

arXiv2026

MolmoSpaces: A Large-Scale Open Ecosystem for Robot Navigation and Manipulation

Deploying robots at scale demands robustness to the long tail of everyday situations. The countless variations in scene layout, object geometry, and task specifications that characterize real environments are vast and underrepresented in existing robot benchmarks. Measuring this level of generalization requires infrastructure at a scale and diversity that physical evaluation alone cannot provide. We introduce MolmoSpaces, a fully open ecosystem to support large-scale benchmarking of robot policies. MolmoSpaces consists of over 230k diverse indoor environments, ranging from handcrafted household scenes to procedurally generated multiroom houses, populated with 130k richly annotated object assets, including 48k manipulable objects with 42M stable grasps. Crucially, these environments are simulator-agnostic, supporting popular options such as MuJoCo, Isaac, and ManiSkill. The ecosystem supports the full spectrum of embodied tasks: static and mobile manipulation, navigation, and multiroom long-horizon tasks requiring coordinated perception, planning, and interaction across entire indoor environments. We also design MolmoSpaces-Bench, a benchmark suite of 8 tasks in which robots interact with our diverse scenes and richly annotated objects. Our experiments show MolmoSpaces-Bench exhibits strong sim-to-real correlation (R = 0.96, \r{ho} = 0.98), confirm newer and stronger zero-shot policies outperform earlier versions in our benchmarks, and identify key sensitivities to prompt phrasing, initial joint positions, and camera occlusion. Through MolmoSpaces and its open-source assets and tooling, we provide a foundation for scalable data generation, policy training, and benchmark creation for robot learning research.

cs.RO↗

arXiv2026

Contact-Anchored Policies: Contact Conditioning Creates Strong Robot Utility Models

The prevalent paradigm in robot learning attempts to generalize across environments, embodiments, and tasks with language prompts at runtime. A fundamental tension limits this approach: language is often too abstract to guide the concrete physical understanding required for robust manipulation. In this work, we introduce Contact-Anchored Policies (CAP), which replace language conditioning with points of physical contact in space. Simultaneously, we structure CAP as a library of modular utility models rather than a monolithic generalist policy. This factorization allows us to implement a real-to-sim iteration cycle: we build EgoGym, a lightweight simulation benchmark, to rapidly identify failure modes and refine our models and datasets prior to real-world deployment. We show that by conditioning on contact and iterating via simulation, CAP generalizes to novel environments and embodiments out of the box on three fundamental manipulation skills while using only 23 hours of demonstration data, and outperforms large, state-of-the-art VLAs in zero-shot evaluations by 56%. All model checkpoints, codebase, hardware, simulation, and datasets will be open-sourced. Project page: https://cap-policy.github.io/

cs.RO↗

arXiv2026

Enhancing Multi-Image Understanding through Delimiter Token Scaling

Large Vision-Language Models (LVLMs) achieve strong performance on single-image tasks, but their performance declines when multiple images are provided as input. One major reason is the cross-image information leakage, where the model struggles to distinguish information across different images. Existing LVLMs already employ delimiter tokens to mark the start and end of each image, yet our analysis reveals that these tokens fail to effectively block cross-image information leakage. To enhance their effectiveness, we propose a method that scales the hidden states of delimiter tokens. This enhances the model's ability to preserve image-specific information by reinforcing intra-image interaction and limiting undesired cross-image interactions. Consequently, the model is better able to distinguish between images and reason over them more accurately. Experiments show performance gains on multi-image benchmarks such as Mantis, MuirBench, MIRB, and QBench2. We further evaluate our method on text-only tasks that require clear distinction. The method improves performance on multi-document and multi-table understanding benchmarks, including TQABench, MultiNews, and WCEP-10. Notably, our method requires no additional training or inference cost.

cs.CV↗

arXiv2026

Knowledge Vector Weakening: Efficient Training-free Unlearning for Large Vision-Language Models

Large Vision-Language Models (LVLMs) are widely adopted for their strong multimodal capabilities, yet they raise serious concerns such as privacy leakage and harmful content generation. Machine unlearning has emerged as a promising solution for removing the influence of specific data from trained models. However, existing approaches largely rely on gradient-based optimization, incurring substantial computational costs for large-scale LVLMs. To address this limitation, we propose Knowledge Vector Weakening (KVW), a training-free unlearning method that directly intervenes in the full model without gradient computation. KVW identifies knowledge vectors that are activated during the model's output generation on the forget set and progressively weakens their contributions, thereby preventing the model from exploiting undesirable knowledge. Experiments on the MLLMU and CLEAR benchmarks demonstrate that KVW achieves a stable forget-retain trade-off while significantly improving computational efficiency over gradient-based and LoRA-based unlearning methods.

cs.LG↗

arXiv2025

Structure-Aware Antibody Design with Affinity-Optimized Inverse Folding

Motivation: The clinical efficacy of antibody therapeutics critically depends on high-affinity target engagement, yet laboratory affinity-maturation campaigns are slow and costly. In computational settings, most protein language models (PLMs) are not trained to favor high-affinity antibodies, and existing preference optimization approaches introduce substantial computational overhead without clear affinity gains. Therefore, this work proposes SimBinder-IF, which converts the inverse folding model ESM-IF into an antibody sequence generator by freezing its structure encoder and training only its decoder to prefer experimentally stronger binders through preference optimization. Results: On the 11-assay AbBiBench benchmark, SimBinder-IF achieves a 55 percent relative improvement in mean Spearman correlation between log-likelihood scores and experimentally measured binding affinity compared to vanilla ESM-IF (from 0.264 to 0.410). In zero-shot generalization across four unseen antigen-antibody complexes, the correlation improves by 156 percent (from 0.115 to 0.294). SimBinder-IF also outperforms baselines in top-10 precision for ten-fold or greater affinity improvements. A case study redesigning antibody F045-092 for A/California/04/2009 (pdmH1N1) shows that SimBinder-IF proposes variants with substantially lower predicted binding free energy changes than ESM-IF (mean Delta Delta G -75.16 vs -46.57). Notably, SimBinder-IF trains only about 18 percent of the parameters of the full ESM-IF model, highlighting its parameter efficiency for high-affinity antibody generation.

cs.CE↗

arXiv2025

Automated Healthcare Thematic Analysis using Multi-Agent Large Language Model: Algorithm Development and Evaluation

Understanding patients experiences is essential for advancing patient-centered care. Qualitative thematic analysis is widely used to explore these experiences, however, the process remains labor-intensive, subjective, and difficult to scale. This study aimed to develop and evaluate Collaborative Theme Identification Agent (CoTI), a multi-agent large language model framework designed to support manual thematic analysis by rapidly generating supporting excerpts, initial codes, and themes. CoTI consists of three agents: Instructor, Thematizer, and CodebookGenerator. The Instructor refines instruction prompts, the Thematizer extracts supporting excerpts and generates initial codes for each transcript, and the CodebookGenerator groups similar codes across all transcripts into a codebook with themes. We evaluated CoTI primarily using 12 heart failure patient transcripts. CoTI-generated outputs were compared against the reference standard developed by senior investigators. To explore human-AI interaction in thematic analysis, we further implemented CoTI in a user-facing application. CoTI generated supporting excerpts, initial codes, and themes that were more similar to those of senior investigators than did the outputs of junior investigators, baseline natural language processing models, and other basic large language models. In an exploratory human-AI collaboration experiment, we found that the collaboration between CoTI and junior investigators provided only marginal gains compared to CoTI alone. A possible hypothesis was that junior investigators may over-rely on CoTI and limit their independent critical thinking. CoTI can improve the efficiency of thematic analysis by rapidly generating supporting excerpts, initial codes, and themes for human researchers review. These findings highlight CoTI potential as a useful tool for scalable qualitative research.

cs.HC↗

arXiv2025

From Time and Place to Preference: LLM-Driven Geo-Temporal Context in Recommendations

Most recommender systems treat timestamps as numeric or cyclical values, overlooking real-world context such as holidays, events, and seasonal patterns. We propose a scalable framework that uses large language models (LLMs) to generate geo-temporal embeddings from only a timestamp and coarse location, capturing holidays, seasonal trends, and local/global events. We then introduce a geo-temporal embedding informativeness test as a lightweight diagnostic, demonstrating on MovieLens, LastFM, and a production dataset that these embeddings provide predictive signal consistent with the outcomes of full model integrations. Geo-temporal embeddings are incorporated into sequential models through (1) direct feature fusion with metadata embeddings or (2) an auxiliary loss that enforces semantic and geo-temporal alignment. Our findings highlight the need for adaptive or hybrid recommendation strategies, and we release a context-enriched MovieLens dataset to support future research.

cs.IR↗

arXiv2025

Chemical States and Local Structure in Cu-Deficient CuInSe2 Thin Films: Insights into Engineering and Bandgap Narrowing

The Cu-deficient CuxInSe2 (x larger than 0.3) phase can be stabilized as a thin film. A uniform Cu-deficient composition with a chalcopyrite structure was obtained by the precision engineering of a two-step synthesis process involving electron-beam evaporation and Se vapor deposition. Detailed structural and chemical analyses were performed employing various X-ray and microscopic techniques to demonstrate that the chemical states and local structure in the Cu-Se-In tetrahedral networks change with the loss of Cu, the In-Se bond becomes shorter, and the In ions become excessively oxidized without phase separation. Moreover, the results indicate that the bandgap narrowing is primarily attributed to the reconstruction of In3+d 5s orbital states. The bandgap narrows from 1.51 eV to 1.4 eV, which is optimal for the photon absorber. Therefore, cation-deficient selenide is promising for stable nontoxic photovoltaics with tunable bandgaps.

cond-mat.mtrl-sci↗

arXiv2025

FinAI Data Assistant: LLM-based Financial Database Query Processing with the OpenAI Function Calling API

We present FinAI Data Assistant, a practical approach for natural-language querying over financial databases that combines large language models (LLMs) with the OpenAI Function Calling API. Rather than synthesizing complete SQL via text-to-SQL, our system routes user requests to a small library of vetted, parameterized queries, trading generative flexibility for reliability, low latency, and cost efficiency. We empirically study three questions: (RQ1) whether LLMs alone can reliably recall or extrapolate time-dependent financial data without external retrieval; (RQ2) how well LLMs map company names to stock ticker symbols; and (RQ3) whether function calling outperforms text-to-SQL for end-to-end database query processing. Across controlled experiments on prices and fundamentals, LLM-only predictions exhibit non-negligible error and show look-ahead bias primarily for stock prices relative to model knowledge cutoffs. Ticker-mapping accuracy is near-perfect for NASDAQ-100 constituents and high for S\&P~500 firms. Finally, FinAI Data Assistant achieves lower latency and cost and higher reliability than a text-to-SQL baseline on our task suite. We discuss design trade-offs, limitations, and avenues for deployment.

cs.IR↗

arXiv2025

GuruAgents: Emulating Wise Investors with Prompt-Guided LLM Agents

This study demonstrates that GuruAgents, prompt-guided AI agents, can systematically operationalize the strategies of legendary investment gurus. We develop five distinct GuruAgents, each designed to emulate an iconic investor, by encoding their distinct philosophies into LLM prompts that integrate financial tools and a deterministic reasoning pipeline. In a backtest on NASDAQ-100 constituents from Q4 2023 to Q2 2025, the GuruAgents exhibit unique behaviors driven by their prompted personas. The Buffett GuruAgent achieves the highest performance, delivering a 42.2\% CAGR that significantly outperforms benchmarks, while other agents show varied results. These findings confirm that prompt engineering can successfully translate the qualitative philosophies of investment gurus into reproducible, quantitative strategies, highlighting a novel direction for automated systematic investing. The source code and data are available at https://github.com/yejining99/GuruAgents.

cs.AI↗

arXiv2025

Improving Fisher Information Estimation and Efficiency for LoRA-based LLM Unlearning

LLMs have demonstrated remarkable performance across various tasks but face challenges related to unintentionally generating outputs containing sensitive information. A straightforward approach to address this issue is to retrain the model after excluding the problematic data. However, this approach incurs prohibitively high computational costs. To overcome this limitation, machine unlearning has emerged as a promising solution that can effectively remove sensitive information without the need to retrain the model from scratch. Recently, FILA has been proposed as a parameter-efficient unlearning method by integrating LoRA adapters. Specifically, it calculates the Fisher information to identify parameters associated with the forget set and assigns them to LoRA adapters for updates. Despite its innovative approach, FILA still requires access to all model parameters and does not adequately account for fundamental assumptions underlying Fisher information, leading to inaccuracies in importance estimation. To address these limitations, we propose VILA, a novel unlearning framework that explicitly considers the assumptions overlooked in FILA, thereby enhancing the accuracy of parameter identification for the forget set. Moreover, VILA significantly reduces computational costs by enabling parameter identification without accessing the entire model. Our method achieves up to 100x higher parameter efficiency and 40x faster training speed compared to FILA, and sets new state-of-the-art performance on benchmarks including TOFU, WMDP, and MUSE. Our code is available at https://github.com/kyj93790/VILA.

cs.LG↗