arXiv subjects

Trung Bui

Publications and source records attributed to Trung Bui.

At least 19 recordsLinked to original sources

arXiv2026

When Attention Closes: How LLMs Lose the Thread in Multi-Turn Interaction

Large language models can follow complex instructions in a single turn, yet over long multi-turn interactions they often lose the thread of instructions, persona, and rules. This degradation has been measured behaviorally but not mechanistically explained. We propose a channel-transition account: goal-defining tokens become less accessible through attention, while goal-related information may persist in residual representations. We introduce the Goal Accessibility Ratio (GAR), measuring attention from generated tokens to task-defining goal tokens, and combine it with sliding-window ablations and residual-stream probes. When attention to instructions closes, what survives reveals architecture. Across architectures, the transition yields qualitatively distinct failure modes: some models preserve goal-conditioned behavior at vanishing attention, others fail despite decodable residual goal information, and the layer at which this encoding emerges varies from 2 to 27. A within-model causal ablation that force-closes the attention channel in Mistral collapses recall from near-perfect to 11% on a 20-fact retention task and raises persona-constraint violations above an adversarial-pressure baseline without user pressure, with both effects emerging at the predictable crossover turn. Linear probes recover per-episode recall outcomes from residual representations with AUC up to 0.99 across all four primary architectures, while input embeddings remain at chance. Across architectures and model scales, the gap between attention loss and residual decodability predicts whether goal-conditioned behavior survives channel closure. We contribute GAR as a diagnostic, the channel-transition framework as a controlled mechanistic account, and a parametric prediction of failure timing under windowed attention closure.

cs.AI↗

arXiv2026

A Survey on LLM-based Conversational User Simulation

User simulation has long played a vital role in computer science due to its potential to support a wide range of applications. Language, as the primary medium of human communication, forms the foundation of social interaction and behavior. Consequently, simulating conversational behavior has become a key area of study. Recent advancements in large language models (LLMs) have significantly catalyzed progress in this domain by enabling high-fidelity generation of synthetic user conversation. In this paper, we survey recent advancements in LLM-based conversational user simulation. We introduce a novel taxonomy covering user granularity and simulation objectives. Additionally, we systematically analyze core techniques and evaluation methodologies. We aim to keep the research community informed of the latest advancements in conversational user simulation and to further facilitate future research by identifying open challenges and organizing existing work under a unified framework.

cs.CL↗

arXiv2026

PageGuide: Browser extension to assist users in navigating a webpage and locating information

Users browsing the web daily struggle to quickly locate relevant information in cluttered pages, and complete multi-step web navigation tasks. State-of-the-art AI assistants (e.g. ChatGPT, Gemini, Claude) and browser agents (e.g. OpenAI Operator, Browser Use) can answer questions and automate actions, yet they return answers without showing where the information comes from on the page, forcing users to manually verify results and blindly trust every automated step. We present PageGuide, a browser extension that grounds LLM answers and actions not only in HTML DOM elements, but also in non-HTML DOM content, such as maps, images, charts, and timetables, addressing two core user needs: (a) FIND: locating and highlighting relevant evidence in-situ so users can instantly verify answers on the page; (b)GUIDE: au tomatically guiding users through navigation tasks while capturing relevant evidence for each intermediate task. In a within-subject user study N=53, PageGuide outperforms a non-grounding agent in both modes: in FIND, it improves accuracy from 90.1% to 96.3% and reduces judge time from 155.6s to 133.0s; in GUIDE, it improves accuracy from 71.4% to 81.0%, with judge time rising from 70.4s to 109.8s as users spend the extra time inspecting grounded evidence across navigation steps rather than believing the answer blindly. Code and demo are at: https://pageguide.github.io/.

cs.HC↗

arXiv2026

SketchVLM: Vision language models can annotate images to explain thoughts and guide users

When answering questions about images, humans naturally point, label, and draw to explain their reasoning. In contrast, modern vision-language models (VLMs) such as Gemini-3-Pro and GPT-5 only respond with text, which can be difficult for users to verify. We present SketchVLM, a training-free, model-agnostic framework that enables VLMs to produce non-destructive, editable SVG overlays on the input image to visually explain their answers. Across seven benchmarks spanning visual reasoning (maze navigation, ball-drop trajectory prediction, and object counting) and drawing (part labeling, connecting-the-dots, and drawing shapes around objects), SketchVLM improves visual reasoning task accuracy by up to +28.5 percentage points and annotation quality by up to 1.48x relative to image-editing and fine-tuned sketching baselines, while also producing annotations that are more faithful to the model's stated answer. We find that single-turn generation already achieves strong accuracy and annotation quality, and multi-turn generation opens up further opportunities for human-AI collaboration. An interactive demo and code are at https://sketchvlm.github.io/.

cs.CV↗

arXiv2026

ViT-AdaLA: Adapting Vision Transformers with Linear Attention

Vision Transformers (ViTs) based vision foundation models (VFMs) have achieved remarkable performance across diverse vision tasks, but suffer from quadratic complexity that limits scalability to long sequences. Existing linear attention approaches for ViTs are typically trained from scratch, requiring substantial computational resources, while linearization-based methods developed for large language model decoders do not transfer well to ViTs. To address these challenges, we propose ViT-AdaLA, a novel framework for effectively adapting and transferring prior knowledge from VFMs to linear attention ViTs. ViT-AdaLA consists of three stages: attention alignment, feature alignment, and supervised fine-tuning. In the attention alignment stage, we align vanilla linear attention with the original softmax-based attention in each block to approximate the behavior of softmax attention. However, residual approximation errors inevitably accumulate across layers. We mitigate this by fine-tuning the linearized ViT to align its final-layer features with a frozen softmax VFM teacher. Finally, the adapted prior knowledge is transferred to downstream tasks through supervised fine-tuning. Extensive experiments on classification and segmentation tasks demonstrate the effectiveness and generality of ViT-AdaLA over various state-of-the-art linear attention counterpart.

cs.CV↗

arXiv2026

ImageEdit-R1: Boosting Multi-Agent Image Editing via Reinforcement Learning

With the rapid advancement of commercial multi-modal models, image editing has garnered significant attention due to its widespread applicability in daily life. Despite impressive progress, existing image editing systems, particularly closed-source or proprietary models, often struggle with complex, indirect, or multi-step user instructions. These limitations hinder their ability to perform nuanced, context-aware edits that align with human intent. In this work, we propose ImageEdit-R1, a multi-agent framework for intelligent image editing that leverages reinforcement learning to coordinate high-level decision-making across a set of specialized, pretrained vision-language and generative agents. Each agent is responsible for distinct capabilities--such as understanding user intent, identifying regions of interest, selecting appropriate editing actions, and synthesizing visual content--while reinforcement learning governs their collaboration to ensure coherent and goal-directed behavior. Unlike existing approaches that rely on monolithic models or hand-crafted pipelines, our method treats image editing as a sequential decision-making problem, enabling dynamic and context-aware editing strategies. Experimental results demonstrate that ImageEdit-R1 consistently outperforms both individual closed-source diffusion models and alternative multi-agent framework baselines across multiple image editing datasets.

cs.CV↗

arXiv2026

Can Large Language Models Keep Up? Benchmarking Online Adaptation to Continual Knowledge Streams

LLMs operating in dynamic real-world contexts often encounter knowledge that evolves continuously or emerges incrementally. To remain accurate and effective, models must adapt to newly arriving information on the fly. We introduce Online Adaptation to Continual Knowledge Streams(OAKS) to evaluate this capability, establishing a benchmark for online adaptation over streaming, continually updating knowledge. Specifically, the benchmark is structured as a sequence of fine-grained context chunks where facts change dynamically across time intervals. OAKS comprises two datasets: OAKS-BABI and OAKS-Novel, where individual facts evolve multiple times across context chunks. These datasets include dense annotations to measure whether models track changes accurately. Evaluating 14 models with varied inference approaches, we observe significant limitations in current methodologies. Both state-of-the-art models and agentic memory systems fail to adapt robustly on OAKS, demonstrating delays in state-tracking and susceptibility to distraction within streaming environments.

cs.CL↗

arXiv2026

Agentic Planning with Reasoning for Image Styling via Offline RL

Direct prompt-based editing often fails on complex transformations because vague and subjective prompts often require nuanced understanding of what should be changed in the image. Our core intuition is that leveraging compositional image editing tools rather than direct prompting profits from structured agent-level planning with explicit reasoning, leading to better results. This structured planning framework enables efficient offline RL post-training on quality-scored trajectories to improve performance. We present a tool-based agentic RL post-training framework that addresses this through structured planning with chain-of-thought reasoning. Our key contributions include: (1) A tool-based agentic planning methodology that combines a compositional library of orthogonal primitive transformations, structured context representation, and explicit per-step reasoning to decompose complex styling into interpretable tool sequences. (2) A synthetic data generation pipeline producing three large-scale datasets (each $\sim$10K trajectories) with reasoning chains, plans, and quality scores, as no existing datasets provide such supervision. Our datasets and code are publicly available at the HuggingFace repository. (3) Offline RL training methods for learning planners with reasoning as our core algorithmic contributions, which consistently improve over the Edit-Only baseline in visual quality and instruction following. (4) Comprehensive evaluation across 4B and 8B parameter Qwen3-VL models showing that our methods outperform other baselines in the majority of compositional tasks, validated by human evaluations.

cs.LG↗

arXiv2026

InfinityStory: Unlimited Video Generation with World Consistency and Character-Aware Shot Transitions

Generating long-form storytelling videos with consistent visual narratives remains a significant challenge in video synthesis. We present a novel framework, dataset, and a model that address three critical limitations: background consistency across shots, seamless multi-subject shot-to-shot transitions, and scalability to hour-long narratives. Our approach introduces a background-consistent generation pipeline that maintains visual coherence across scenes while preserving character identity and spatial relationships. We further propose a transition-aware video synthesis module that generates smooth shot transitions for complex scenarios involving multiple subjects entering or exiting frames, going beyond the single-subject limitations of prior work. To support this, we contribute with a synthetic dataset of 10,000 multi-subject transition sequences covering underrepresented dynamic scene compositions. On VBench, InfinityStory achieves the highest Background Consistency (88.94), highest Subject Consistency (82.11), and the best overall average rank (2.80), showing improved stability, smoother transitions, and better temporal coherence.

cs.CV↗

arXiv2025

StreamGaze: Gaze-Guided Temporal Reasoning and Proactive Understanding in Streaming Videos

Streaming video understanding requires models not only to process temporally incoming frames, but also to anticipate user intention for realistic applications such as Augmented Reality (AR) glasses. While prior streaming benchmarks evaluate temporal reasoning, none measure whether Multimodal Large Language Models (MLLMs) can interpret or leverage human gaze signals within a streaming setting. To fill this gap, we introduce StreamGaze, the first benchmark designed to evaluate how effectively MLLMs utilize gaze for temporal and proactive reasoning in streaming videos. StreamGaze introduces gaze-guided past, present, and proactive tasks that comprehensively assess streaming video understanding. These tasks evaluate whether models can use real-time gaze signals to follow shifting attention and infer user intentions based only on past and currently observed frames. To build StreamGaze, we develop a gaze-video Question Answering (QA) generation pipeline that aligns egocentric videos with raw gaze trajectories through fixation extraction, region-specific visual prompting, and scanpath construction. This pipeline produces spatio-temporally grounded QA pairs that reflect human perceptual dynamics. Across all StreamGaze tasks, we observe substantial performance gaps between state-of-the-art MLLMs and human performance, highlighting key limitations in gaze-based temporal reasoning, intention modeling, and proactive prediction. We further provide detailed analyses of gaze prompting strategies, reasoning behaviors, and task-specific failure modes, offering insights into current limitations and directions for future research. All data and code are publicly available to support continued research in gaze-guided streaming video understanding.

cs.CV↗

arXiv2025

Drift No More? Context Equilibria in Multi-Turn LLM Interactions

Large Language Models (LLMs) excel at single-turn tasks such as instruction following and summarization, yet real-world deployments require sustained multi-turn interactions where user goals and conversational context persist and evolve. A recurring challenge in this setting is context drift: the gradual divergence of a model's outputs from goal-consistent behavior across turns. Unlike single-turn errors, drift unfolds temporally and is poorly captured by static evaluation metrics. In this work, we present a study of context drift in multi-turn interactions and propose a simple dynamical framework to interpret its behavior. We formalize drift as the turn-wise KL divergence between the token-level predictive distributions of the test model and a goal-consistent reference model, and propose a recurrence model that interprets its evolution as a bounded stochastic process with restoring forces and controllable interventions. We instantiate this framework in both synthetic long-horizon rewriting tasks and realistic user-agent simulations such as in $\tau$-Bench, measuring drift for several open-weight LLMs that are used as user simulators. Our experiments consistently reveal stable, noise-limited equilibria rather than runaway degradation, and demonstrate that simple reminder interventions reliably reduce divergence in line with theoretical predictions. Together, these results suggest that multi-turn drift can be understood as a controllable equilibrium phenomenon rather than as inevitable decay, providing a foundation for studying and mitigating context drift in extended interactions.

cs.CL↗

arXiv2025

Steering MoE LLMs via Expert (De)Activation

Mixture-of-Experts (MoE) in Large Language Models (LLMs) routes each token through a subset of specialized Feed-Forward Networks (FFN), known as experts. We present SteerMoE, a framework to steer MoE models by detecting and controlling behavior-associated experts. We detect key experts by comparing how often they activate between paired inputs that demonstrate opposite behaviors (e.g., safe vs. unsafe). By selectively activating or deactivating such experts during inference, we control behaviors like faithfulness and safety without fine-tuning. Across 11 benchmarks and 6 LLMs, our steering raises safety by up to +20% and faithfulness by +27%. Alternatively, unsafe steering drops safety by -41% alone, and -100% when combined with existing jailbreak methods, bypassing all safety guardrails. Overall, SteerMoE offers a lightweight, effective, and widely applicable test-time control, while revealing unique vulnerabilities in MoE LLMs. https://github.com/adobe-research/SteerMoE

cs.CL↗

arXiv2025

Lizard: An Efficient Linearization Framework for Large Language Models

We propose Lizard, a linearization framework that transforms pretrained Transformer-based Large Language Models (LLMs) into subquadratic architectures. Transformers faces severe computational and memory bottlenecks with long sequences due to the quadratic complexity of softmax attention and the growing Key-Value (KV) cache that makes inference memory-bound by context length. Lizard addresses these limitations by introducing a subquadratic attention mechanism that closely approximates softmax attention while preserving model quality. Unlike prior linearization methods constrained by fixed, non-adaptive structures, Lizard augments the architecture with compact, learnable modules that enable adaptive memory control and robust length generalization. Moreover, we introduce a hardwareaware algorithm that solves numerical instability in gated attention to accelerate training. Extensive experiments show that Lizard achieves near-lossless recovery of its teacher model's performance, significantly outperforming previous methods by up to 9.4 - 24.5 points on the 5-shot MMLU benchmark and demonstrating superior associative recall.

cs.CL↗

arXiv2025

FIFA: Unified Faithfulness Evaluation Framework for Text-to-Video and Video-to-Text Generation

Video Multimodal Large Language Models (VideoMLLMs) have achieved remarkable progress in both Video-to-Text and Text-to-Video tasks. However, they often suffer fro hallucinations, generating content that contradicts the visual input. Existing evaluation methods are limited to one task (e.g., V2T) and also fail to assess hallucinations in open-ended, free-form responses. To address this gap, we propose FIFA, a unified FaIthFulness evAluation framework that extracts comprehensive descriptive facts, models their semantic dependencies via a Spatio-Temporal Semantic Dependency Graph, and verifies them using VideoQA models. We further introduce Post-Correction, a tool-based correction framework that revises hallucinated content. Extensive experiments demonstrate that FIFA aligns more closely with human judgment than existing evaluation methods, and that Post-Correction effectively improves factual consistency in both text and video generation.

cs.CV↗

arXiv2025

Instruction Tuning with and without Context: Behavioral Shifts and Downstream Impact

Instruction tuning is a widely used approach to improve the instruction-following ability of large language models (LLMs). Instruction-tuning datasets typically include a mixture of context-augmented and context-free examples, yet prior work has largely combined these data types without examining their distinct effects. In this paper, we investigate how training LLMs with or without context affects model behavior and downstream performance. First, in the text domain, we show that LLMs trained with context attend more strongly to the provided knowledge, achieving better grounding. We also observe that context-augmented training shifts how LLMs use knowledge: models store and leverage less on parametric knowledge and instead depend more on the provided context. Second, we observe that using LLM trained with context-augmented data as the backbone for vision-language models reduces hallucination and improves grounding in the visual domain. Finally, we explore practical strategies for real-world deployments where context availability varies. We show that maintaining separate context-augmented and context-free models and routing inputs between them yields more robust overall performance than training a single mixed model, as it better preserves their complementary strengths.

cs.CL↗

arXiv2025

MS4UI: A Dataset for Multi-modal Summarization of User Interface Instructional Videos

We study multi-modal summarization for instructional videos, whose goal is to provide users an efficient way to learn skills in the form of text instructions and key video frames. We observe that existing benchmarks focus on generic semantic-level video summarization, and are not suitable for providing step-by-step executable instructions and illustrations, both of which are crucial for instructional videos. We propose a novel benchmark for user interface (UI) instructional video summarization to fill the gap. We collect a dataset of 2,413 UI instructional videos, which spans over 167 hours. These videos are manually annotated for video segmentation, text summarization, and video summarization, which enable the comprehensive evaluations for concise and executable video summarization. We conduct extensive experiments on our collected MS4UI dataset, which suggest that state-of-the-art multi-modal summarization methods struggle on UI video summarization, and highlight the importance of new methods for UI instructional video summarization.

cs.CV↗

arXiv2025

Offline RL by Reward-Weighted Fine-Tuning for Conversation Optimization

Offline reinforcement learning (RL) is a variant of RL where the policy is learned from a previously collected dataset of trajectories and rewards. In our work, we propose a practical approach to offline RL with large language models (LLMs). We recast the problem as reward-weighted fine-tuning, which can be solved using similar techniques to supervised fine-tuning (SFT). To showcase the value of our approach, we apply it to learning short-horizon question-answering policies of a fixed length, where the agent reasons about potential answers or asks clarifying questions. Our work stands in a stark contrast to state-of-the-art methods in this domain, based on SFT and direct preference optimization, which have additional hyper-parameters and do not directly optimize for rewards. We compare to them empirically, and report major gains in both optimized rewards and language quality.

cs.CL↗

arXiv2025

Understanding Generative AI Capabilities in Everyday Image Editing Tasks

Generative AI (GenAI) holds significant promise for automating everyday image editing tasks, especially following the recent release of GPT-4o on March 25, 2025. However, what subjects do people most often want edited? What kinds of editing actions do they want to perform (e.g., removing or stylizing the subject)? Do people prefer precise edits with predictable outcomes or highly creative ones? By understanding the characteristics of real-world requests and the corresponding edits made by freelance photo-editing wizards, can we draw lessons for improving AI-based editors and determine which types of requests can currently be handled successfully by AI editors? In this paper, we present a unique study addressing these questions by analyzing 83k requests from the past 12 years (2013-2025) on the Reddit community, which collected 305k PSR-wizard edits. According to human ratings, approximately only 33% of requests can be fulfilled by the best AI editors (including GPT-4o, Gemini-2.0-Flash, SeedEdit). Interestingly, AI editors perform worse on low-creativity requests that require precise editing than on more open-ended tasks. They often struggle to preserve the identity of people and animals, and frequently make non-requested touch-ups. On the other side of the table, VLM judges (e.g., o1) perform differently from human judges and may prefer AI edits more than human edits. Code and qualitative examples are available at: https://psrdataset.github.io

cs.CV↗