When Users Change Their Mind: Evaluating Interruptible Agents in Long-Horizon Web Navigation

As LLM agents transition from short, static problem solving to executing complex, long-horizon tasks in dynamic environments, the ability to handle user interruptions, such as adding requirement or revising goals, during mid-task execution is becoming a core requirement for realistic deployment. However, existing benchmarks largely assume uninterrupted agent behavior or study interruptions only in short, unconstrained language tasks. In this paper, we present the first systematic study of interruptible agents in long-horizon, environmentally grounded web navigation tasks, where actions induce persistent state changes. We formalize three realistic interruption types, including addition, revision, and retraction, and introduce InterruptBench, a benchmark derived from WebArena-Lite that synthesizes high-quality interruption scenarios under strict semantic constraints. Using a unified interruption simulation framework, we evaluate six strong LLM backbones across single- and multi-turn interruption settings, analyzing both their effectiveness in adapting to updated intents and their efficiency in recovering from mid-task changes. Our results show that handling user interruptions effectively and efficiently during long-horizon agentic tasks remains challenging for powerful large-scale LLMs. Code and dataset are available at https://github.com/HenryPengZou/InterruptBench.

Cold-Starts in Generative Recommendation: A Reproducibility Study

Cold-start recommendation remains a central challenge in dynamic, open-world platforms, requiring models to recommend for newly registered users (user cold-start) and to recommend newly introduced items to existing users (item cold-start) under sparse or missing interaction signals. Recent generative recommenders built on pre-trained language models (PLMs) are often expected to mitigate cold-start by using item semantic information (e.g., titles and descriptions) and test-time conditioning on limited user context. However, cold-start is rarely treated as a primary evaluation setting in existing studies, and reported gains are difficult to interpret because key design choices, such as model scale, identifier design, and training strategy, are frequently changed together. In this work, we present a systematic reproducibility study of generative recommendation under a unified suite of cold-start protocols.

MathGen: Revealing the Illusion of Mathematical Competence through Text-to-Image Generation

Modern generative models have demonstrated the ability to solve challenging mathematical problems. In many real-world settings, however, mathematical solutions must be expressed visually through diagrams, plots, geometric constructions, and structured symbolic layouts, where correctness depends on precise visual composition. This naturally raises the question of whether generative models can still do so when the answer must be rendered visually rather than written in text? To study this problem, we introduce MathGen, a rigorous benchmark of 900 problems spanning seven core domains, each paired with an executable verifier under a Script-as-a-Judge protocol for deterministic and objective evaluation. Experiments on representative open-source and proprietary text-to-image models show that mathematical fidelity remains a major bottleneck: even the best closed-source model reaches only 42.0% overall accuracy, while open-source models achieve just ~ 1-11%, often near 0% on structured tasks. Overall, current T2I models remain far from competent at even elementary mathematical visual generation.

MiroEval: Benchmarking Multimodal Deep Research Agents in Process and Outcome

Recent progress in deep research systems has been impressive, but evaluation still lags behind real user needs. Existing benchmarks predominantly assess final reports using fixed rubrics, failing to evaluate the underlying research process. Most also offer limited multimodal coverage, rely on synthetic tasks that do not reflect real-world query complexity, and cannot be refreshed as knowledge evolves. To address these gaps, we introduce MiroEval, a benchmark and evaluation framework for deep research systems. The benchmark comprises 100 tasks (70 text-only, 30 multimodal), all grounded in real user needs and constructed via a dual-path pipeline that supports periodic updates, enabling a live and evolving setting. The proposed evaluation suite assesses deep research systems along three complementary dimensions: adaptive synthesis quality evaluation with task-specific rubrics, agentic factuality verification via active retrieval and reasoning over both web sources and multimodal attachments, and process-centric evaluation audits how the system searches, reasons, and refines throughout its investigation. Evaluation across 13 systems yields three principal findings: the three evaluation dimensions capture complementary aspects of system capability, with each revealing distinct strengths and weaknesses across systems; process quality serves as a reliable predictor of overall outcome while revealing weaknesses invisible to output-level metrics; and multimodal tasks pose substantially greater challenges, with most systems declining by 3 to 10 points. The MiroThinker series achieves the most balanced performance, with MiroThinker-H1 ranking the highest overall in both settings. Human verification and robustness results confirm the reliability of the benchmark and evaluation framework. MiroEval provides a holistic diagnostic tool for the next generation of deep research agents.

AgentSwing: Adaptive Parallel Context Management Routing for Long-Horizon Web Agents

As large language models (LLMs) evolve into autonomous agents for long-horizon information-seeking, managing finite context capacity has become a critical bottleneck. Existing context management methods typically commit to a single fixed strategy throughout the entire trajectory. Such static designs may work well in some states, but they cannot adapt as the usefulness and reliability of the accumulated context evolve during long-horizon search. To formalize this challenge, we introduce a probabilistic framework that characterizes long-horizon success through two complementary dimensions: search efficiency and terminal precision. Building on this perspective, we propose AgentSwing, a state-aware adaptive parallel context management routing framework. At each trigger point, AgentSwing expands multiple context-managed branches in parallel and uses lookahead routing to select the most promising continuation. Experiments across diverse benchmarks and agent backbones show that AgentSwing consistently outperforms strong static context management methods, often matching or exceeding their performance with up to $3\times$ fewer interaction turns while also improving the ultimate performance ceiling of long-horizon web agents. Beyond the empirical gains, the proposed probabilistic framework provides a principled lens for analyzing and designing future context management strategies for long-horizon agents.

VC-Soup: Value-Consistency Guided Multi-Value Alignment for Large Language Models

As large language models (LLMs) increasingly shape content generation, interaction, and decision-making across the Web, aligning them with human values has become a central objective in trustworthy AI. This challenge becomes even more pronounced when aligning multiple, potentially conflicting human values. Although recent approaches, such as reward reweighting, prompt-based supervised fine-tuning, and model merging, attempt to tackle multi-value alignment, they still face two major limitations: (1) training separate models for each value combination is prohibitively expensive; (2) value conflicts substantially degrade alignment performance. These limitations make it difficult to achieve favorable trade-offs across diverse human values. To address these challenges, we revisit multi-value alignment from the perspective of value consistency in data and propose VC-soup, a data filtering and parameter merging framework grounded in value-consistent learning. We first design a value consistency metric based on the cosine similarity between the reward-gap vector of each preference pair and an all-ones vector, which quantifies its cross-value coherence. We then filter out low-consistency preference pairs in each value dataset and train on the remaining data to obtain smooth, value-consistent policy models that better preserve linear mode connectivity. Finally, we linearly combine these policies and apply Pareto filtering across values to obtain solutions with balanced multi-value performance. Extensive experiments and theoretical analysis demonstrate that VC-soup effectively mitigates conflicts and consistently outperforms existing multi-value alignment methods.

Chem4DLLM: 4D Multimodal LLMs for Chemical Dynamics Understanding

Existing chemical understanding tasks primarily rely on static molecular representations, limiting their ability to model inherently dynamic phenomena such as bond breaking or conformational changes, which are essential for a chemist to understand chemical reactions. To address this gap, we introduce Chemical Dynamics Understanding (ChemDU), a new task that translates 4D molecular trajectories into interpretable natural-language explanations. ChemDU focuses on fundamental dynamic scenarios, including gas-phase and catalytic reactions, and requires models to reason about key events along molecular trajectories, such as bond formation and dissociation, and to generate coherent, mechanistically grounded narratives. To benchmark this capability, we construct Chem4DBench, the first dataset pairing 4D molecular trajectories with expert-authored explanations across these settings. We further propose Chem4DLLM, a unified model that integrates an equivariant graph encoder with a pretrained large language model to explicitly capture molecular geometry and rotational dynamics. We hope that ChemDU, together with Chem4DBench and Chem4DLLM, will stimulate further research in dynamic chemical understanding and multimodal scientific reasoning.

Causally Sufficient and Necessary Feature Expansion for Class-Incremental Learning

Current expansion-based methods for Class Incremental Learning (CIL) effectively mitigate catastrophic forgetting by freezing old features. However, such task-specific features learned from the new task may collide with the old features. From a causal perspective, spurious feature correlations are the main cause of this collision, manifesting in two scopes: (i) guided by empirical risk minimization (ERM), intra-task spurious correlations cause task-specific features to rely on shortcut features. These non-robust features are vulnerable to interference, inevitably drifting into the feature space of other tasks; (ii) inter-task spurious correlations induce semantic confusion between visually similar classes across tasks. To address this, we propose a Probability of Necessity and Sufficiency (PNS)-based regularization method to guide feature expansion in CIL. Specifically, we first extend the definition of PNS to expansion-based CIL, termed CPNS, which quantifies both the causal completeness of intra-task representations and the separability of inter-task representations. We then introduce a dual-scope counterfactual generator based on twin networks to ensure the measurement of CPNS, which simultaneously generates: (i) intra-task counterfactual features to minimize intra-task PNS risk and ensure causal completeness of task-specific features, and (ii) inter-task interfering features to minimize inter-task PNS risk, ensuring the separability of inter-task representations. Theoretical analyses confirm its reliability. The regularization is a plug-and-play method for expansion-based CIL to mitigate feature collision. Extensive experiments demonstrate the effectiveness of the proposed method.

Model Editing for New Document Integration in Generative Information Retrieval

Generative retrieval (GR) reformulates the Information Retrieval (IR) task as the generation of document identifiers (docIDs). Despite its promise, existing GR models exhibit poor generalization to newly added documents, often failing to generate the correct docIDs. While incremental training offers a straightforward remedy, it is computationally expensive, resource-intensive, and prone to catastrophic forgetting, thereby limiting the scalability and practicality of GR. In this paper, we identify the core bottleneck as the decoder's ability to map hidden states to the correct docIDs of newly added documents. Model editing, which enables targeted parameter modifications for docID mapping, represents a promising solution. However, applying model editing to current GR models is not trivial, which is severely hindered by indistinguishable edit vectors across queries, due to the high overlap of shared docIDs in retrieval results. To address this, we propose DOME (docID-oriented model editing), a novel method that effectively and efficiently adapts GR models to unseen documents. DOME comprises three stages: (1) identification of critical layers, (2) optimization of edit vectors, and (3) construction and application of updates. At its core, DOME employs a hybrid-label adaptive training strategy that learns discriminative edit vectors by combining soft labels, which preserve query-specific semantics for distinguishable updates, with hard labels that enforce precise mapping modifications. Experiments on widely used benchmarks, including NQ and MS MARCO, show that our method significantly improves retrieval performance on new documents while maintaining effectiveness on the original collection. Moreover, DOME achieves this with only about 60% of the training time required by incremental training, considerably reducing computational cost and enabling efficient, frequent model updates.

Truth as a Trajectory: What Internal Representations Reveal About Large Language Model Reasoning

Existing explainability methods for Large Language Models (LLMs) typically treat hidden states as static points in activation space, assuming that correct and incorrect inferences can be separated using representations from an individual layer. However, these activations are saturated with polysemantic features, leading to linear probes learning surface-level lexical patterns rather than underlying reasoning structures. We introduce Truth as a Trajectory (TaT), which models the transformer inference as an unfolded trajectory of iterative refinements, shifting analysis from static activations to layer-wise geometric displacement. By analyzing displacement of representations across layers, TaT uncovers geometric invariants that distinguish valid reasoning from spurious behavior. We evaluate TaT across dense and Mixture-of-Experts (MoE) architectures on benchmarks spanning commonsense reasoning, question answering, and toxicity detection. Without access to the activations themselves and using only changes in activations across layers, we show that TaT effectively mitigates reliance on static lexical confounds, outperforming conventional probing, and establishes trajectory analysis as a complementary perspective on LLM explainability.