NeurIPS 2026 论文列表

While recent multimodal recommender systems have demonstrated the effectiveness of incorporating visual and textual information to improve downstream performance, most existing methods rely on static modality fusion, assuming that the relative importance of textual and visual signals remains stable across recommendation scenarios. This design may not fully account for an important variation across recommendation requests: some queries require fine-grained visual cues, whereas others are better served by textual or functional semantics, in which case indiscriminate modality fusion brings in uninformative cues and impairs recommendation quality. To address this, we propose AdaM-Rec, an LLM-based framework for adaptive modality routing in multimodal recommendation, which enables dynamic calibration of reliance on textual and multimodal evidence for user-specific queries. Built on structured natural-language representations of items and user preferences, it estimates modality reliability using proxy recall tasks. Specifically, it generates pseudo-queries that match the granularity of the actual query while pointing to the user's positively interacted items as verifiable proxy targets, evaluating which modality yields better recall performance in analogous scenarios and optimizing the routing strategy in an agentic manner. It then performs routed recall with optimized strategy, enriches results with collaborative items, and ranks candidates by their relevance to both the query and user preferences. Experiments demonstrate that AdaM-Rec delivers strong performance against state-of-the-art baselines, highlighting the effectiveness and broader potential of adaptive control over modality reliance in multimodal recommendation. Code will be publicly available.

Large Language Models (LLMs) deployed as AI agents frequently exhibit a task-completion bias, executing hallucinated, undesired actions to force a resolution rather than expressing uncertainty. Existing detection methods fail to provide actionable, real-time correction as they either do not localize the hallucinations or incur prohibitive inference latency. We introduce the Latent Critic, a lightweight low-rank adapter (LoRA) that operates concurrently with a frozen base LLM's generation to actively restructure the transformer's residual stream---amplifying epistemic uncertainty signals and translating them into localized, natural language feedback within a single sequence. By refining the base model's native uncertainty signals, this manipulation of the latent space enables highly reliable, granular detection without the overhead of secondary inference loops. Mechanistic analysis via activation patching and layer-wise probing shows that this rank-invariant behavior restructures pre-existing uncertainty geometry into a linearly separable representation. Using tool-calling as an instantiation of granular hallucinations, we validate the detection and downstream improvements enabled by the Latent Critic architecture across Qwen and Llama-based models. Demonstrating superior real-time efficacy, our approach significantly outperforms equivalent-scale external detectors and internal probes in isolating hallucinations (0.870 vs. 0.695 F1), achieving >80% accuracy in localization (e.g., ungrounded: date). When deployed in a closed-loop ReAct environment, the Critic acts as a zero-latency guardrail, intercepting hallucinations before execution to prevent undesired actions while simultaneously enabling efficient agent self-correction.

Professional image retouching is a sequential and reasoning-driven process. Photographers first analyze scene semantics, lighting structure, and depth cues to design a retouching plan, and then progressively apply global tone grading and mask-guided local corrections based on the evolving visual outcome. Recent MLLM-empowered retouching agents either predict global parameters or respond to user-provided retouching instructions. Although showing impressive results, they lack the capacity for autonomous sequential reasoning and fall short of fine-grained local control. We propose **AURA**, an **AU**tonomous **R**etouching **A**gent that emulates the workflow of expert photographers. Given an input photograph, AURA autonomously assesses the scene and proceeds progressively. At each step, it reasons about what to retouch, executes the adjustment, and feeds the result back as visual context for next-step reasoning, forming a fine-grained visual chain-of-thought that evolves with the image. To support AURA, we contribute the first expert-annotated long-horizon retouching trajectory dataset, where expert photographers edit the input image from scratch with deliberate artistic intent. Based on this dataset, we train AURA in two stages: supervised fine-tuning to learn the retouching trajectory, followed by GRPO-retouching, an agentic reinforcement learning stage that optimizes for perceptual quality and tool accuracy. Our experiments demonstrate that AURA consistently outperforms existing methods in both quantitative metrics and user studies, enabling one-click professional retouching and democratizing masterpiece previously accessible only to skilled experts.

Visual reasoning, often interleaved with intermediate visual states, has emerged as a promising direction in the field. A straightforward approach is to directly generate images via unified models during reasoning, but this is computationally expensive and architecturally non-trivial. Recent alternatives include agentic reasoning through code or tool calls, and latent reasoning with learnable hidden embeddings. However, agentic methods incur context-switching latency from external execution, while latent methods lack task generalization and are difficult to train with autoregressive parallelization. To combine their strengths while mitigating their limitations, we propose ATLAS, a framework in which a single discrete "word", termed as a functional token, serves both as an agentic operation and a latent visual reasoning unit. Each functional token is associated with an internalized visual operation, yet remains a standard token in the tokenizer vocabulary and can be generated via next-token prediction. This design avoids verbose intermediate visual content generation, while preserving compatibility with standard SFT and RL training, without architectural or methodological modifications. To further address the sparsity of functional tokens during RL, we introduce Latent-Anchored GRPO (LA-GRPO), which stabilizes the training by anchoring functional tokens with a statically weighted auxiliary objective, providing stronger gradient updates. Extensive experiments and analyses demonstrate that ATLAS achieves superior performance on challenging benchmarks while maintaining clear interpretability. We hope ATLAS offers a new paradigm inspiring future visual reasoning research.

Clarification-seeking behavior is widely regarded as a desirable property of LLM agents, enabling them to resolve ambiguity before acting on underspecified tasks. However, the security implications of this interaction pattern remain unexplored. We investigate whether the transition from standard execution to a clarification-seeking state increases an agent's susceptibility to prompt injection attacks. We introduce **ASPI** (**A**mbiguous-**S**tate **P**rompt **I**njection), a benchmark of 728 task-attack scenarios that isolates clarification as a distinct agent state and measures how this state transition affects vulnerability under controlled conditions. Each benchmark instance is evaluated under matched execution and clarification settings: in the execution setting, the agent acts on a fully specified instruction and encounters adversarial content only through tool-returned data; in the clarification setting, the agent must first request and incorporate additional user input before acting. We evaluate ten frontier LLMs and find that clarification-seeking consistently and substantially amplifies vulnerability. For instance, attack success rises from 1.8\% to 34.0\% for o3 and from 2.2\% to 35.7\% for Gemini-3-Flash. A decomposition analysis reveals that this gap reflects both a state-dependent shift in how models process incoming content and a channel-specific effect arising from the agent-solicited clarification interface. These findings demonstrate that standard execution-time security evaluation systematically underestimates the attack surface of interactive agents, and that robustness under fully specified tasks does not translate to robustness under ambiguity. For reproducibility our data and source code is available at https://anonymous.4open.science/r/aspi-5074.

Long-horizon embodied tasks remain a fundamental challenge in AI, as current methods rely on hand-engineered rewards or action-labeled demonstrations, neither of which scales. We introduce ASH, an agentic system that learns an embodied policy from unlabeled, noisy internet video, without reward shaping or expert annotation. ASH follows a self-improvement loop; when it gets stuck, ASH learns an Inverse Dynamics Model (IDM) from its own trajectories, and uses its IDM to extract supervision from relevant internet video. ASH uses unsupervised learning to identify key moments from large-scale internet video and retains them as long-term memory --- allowing it to tackle long-horizon problems. We evaluate ASH on two complementary environments demanding multi-hour planning: Pokemon Emerald, a turn-based RPG, and The Legend of Zelda: The Minish Cap, a real-time action-adventure game. In both games, behavioral cloning, retrieval-augmented and zero-shot foundation-model baselines plateau, while ASH sustains progression across our 8-hour evaluation. ASH reaches an average of $11.2/12$ milestones in Pokemon Emerald and $9.9/12$ in Legend of Zelda, while the strongest baseline gets stuck in both environments at an average of $6.5/12$ and $6.0/12$ milestones, respectively. We demonstrate that self-improving agents are a scalable recipe for long-horizon embodied learning.

Current test-time scaling (TTS) techniques enhance large language model (LLM) performance by allocating additional computation at inference time, yet they remain insufficient for agentic settings, where actions directly interact with external environments and their effects can be irreversible and costly. We propose ARTIS, Agentic Risk-Aware Test-Time Scaling via Iterative Simulation, a framework that decouples exploration from commitment by enabling test-time exploration through simulated interactions prior to real-world execution. This design allows extending inference-time computation to improve action-level reliability and robustness without incurring environmental risk. We further show that naive LLM-based simulators struggle to capture rare but high-impact failure modes, substantially limiting their effectiveness for agentic decision making. To address this limitation, we introduce a risk-aware tool simulator that emphasizes fidelity on failure-inducing actions via targeted data generation and rebalanced training. Experiments on multi-turn and multi-step agentic benchmarks demonstrate that iterative simulation substantially improves agent reliability, and that risk-aware simulation is essential for consistently realizing these gains across models and tasks.

Existing multimodal retrieval benchmarks largely emphasize semantic matching on daily-life images and offer limited diagnostics of professional knowledge and complex reasoning. To address this gap, we introduce ARK, a benchmark designed to analyze multimodal retrieval from two complementary perspectives: (i) knowledge domains (five domains with 17 subtypes), which characterize the content and expertise retrieval relies on, and (ii) reasoning skills (six categories), which characterize the type of inference over multimodal evidence required to identify the correct candidate. Specifically, ARK evaluates retrieval with both unimodal and multimodal queries and candidates, covering 16 heterogeneous visual data types. To avoid shortcut matching during evaluation, most queries are paired with targeted hard negatives that require multi-step reasoning. We evaluate 26 representative text-based and multimodal retrievers and observe a pronounced gap between knowledge- and reasoning-intensive retrieval, with fine-grained visual and spatial reasoning as persistent bottlenecks. We further show that enhancements such as re-ranking, rewriting, and agentic retrieval yield consistent gains, but substantial headroom remains. The dataset and evaluation pipeline will be released soon.

We introduce AIRA-Compose — a framework that enables AI research agents to autonomously explore and discover novel neural architectures. AIRA-Compose leverages the previously-proposed Composer abstraction to define a combinatorial design space over fundamental model primitives (Attention, MLP, and Mamba), then uses a guided search policy to efficiently navigate this space for novel neural architecture design. Our approach operates in two stages: (1) agents iteratively designs and evaluates candidate architectures at the million-parameter scale, then (2) top-performing candidates are scaled to billion parameters for pre-training. Overall, AIRA-Compose discovers 14 novel architectures spanning two families — AIRAformers (Transformer-based) and AIRAhybrids (Transformer-Mamba Hybrid-based). When pre-trained at 1B scale under a fixed token budget, agent-discovered top-performing architectures consistently outperform both Llama 3.2 and Composer-found alternatives. On downstream tasks, AIRAformer-D and AIRAhybrid-D improve accuracy by 2.4% and 3.8% over Llama 3.2, respectively. AIRA-Compose also finds novel model architectures that achieve steeper, more efficient compute-optimal scaling frontiers. AIRAformer-C scales 54% and 71% faster than Llama 3.2 and the best Composer-found Transformer, while AIRAhybrid-C scales 23% and 37% faster than the modified Nemotron-2 and the best Composer-found hybrid, respectively. For the first time, AIRA-Compose demonstrates that AI research agents can discover hybrid architectures that surpass hand-designed baselines on the accuracy–compute frontier. It establishes a flexible paradigm for LLMs to discover next generation foundation models, marking a step towards recursive self-improvement.

Existing research has identified three structural performance bottlenecks in AI research agents: (1) synchronous single-GPU execution constrains sample throughput, limiting the benefit of search; (2) a generalization gap where validation-based selection causes overfitting and performance to degrade over extended search horizons; and (3) the limited capability of fixed, single-turn LLM operators imposes a ceiling on search performance. We introduce AIRA$_2$, which addresses these bottlenecks through three architectural choices: an asynchronous multi-GPU worker pool that increases experiment throughput linearly; a Hidden Consistent Evaluation protocol that delivers a reliable evaluation signal; and ReAct agents that dynamically scope their actions and debug interactively. On MLE-bench-30, AIRA$_2$ achieves a mean Percentile Rank of 81.5% at 24 hours and 83.1% at 72 hours, outperforming the strongest baseline, which achieves 72.7%. On AIRS-Bench, AIRA$_2$ exceeds human state-of-the-art on 6 out of 20 diverse research tasks. Ablations confirm that each architectural component is necessary, that performance follows a predictable scaling law that transfers across LLM backbones, and that the "overfitting" reported in prior work was driven by evaluation noise rather than true data memorization.

Multi-Agent Debate (MAD) improves reasoning on multimodal tasks, but its accuracy and token efficiency depend on how agents interact during debate. A largely overlooked design choice is which interaction modalities agents use to exchange evidence. Existing methods rely on fixed text- or graph-based interaction, which fails to convey fine-grained multimodal evidence, degrading accuracy while inflating token cost due to verbose descriptions. To address this, we propose Adaptive Interaction in Multi-Agent Debate (AIM), an adaptive framework that dynamically selects the best combination of interaction modalities for each instance. Beyond text and graph, AIM introduces task-dependent modality interaction, where agents exchange targeted regions of interest from the task-specific multimodal input, such as image regions, audio segments, and spatio-temporal clips. To determine which combination is the best for each instance, AIM first generates a structured baseline response, extracts interpretable routing features, and finally uses a lightweight router to select a modality combination that avoids both under-fusion (i.e., agents miss decisive evidence) and over-fusion (i.e., redundant modalities inflate token cost) risks. Across eight multimodal question answering (QA) benchmarks of Visual-QA, Audio-QA, and Video-QA, AIM achieves the highest accuracy on all datasets, improving upon the best single-modality baseline by up to 8.9% while reducing token cost by up to 53.4% relative to the best fixed-modality baseline.

In this paper, we consider mixed-motive games (MMGs) which model various real-world multi-agent scenarios where self-interested agents must both cooperate and compete. We identify an issue with existing gifting methods, where agents become overly dependent on incoming gifts and fail to learn reward-yielding behaviors, a phenomenon we call *gift over-reliance*. To address this, we propose a refusal-augmented gifting framework that balances cooperation and competition in MMGs. Our approach introduces an adaptive refusal mechanism that allows agents to selectively reject a portion of received gifts, thereby mitigating gift over-reliance and preventing the agents from being exploited by other agents. This mechanism is gifting-method-agnostic, serving as a plug-in for existing schemes while enabling decentralized learning. We demonstrate that across diverse MMG environments and gifting schemes, our mechanism consistently reduces gift over-reliance, prevents lock-in to sacrificing policies, and improves $\alpha$-fairness without necessitating major changes to existing training pipelines.

Reinforcement learning (RL) has substantially improved the ability of large language model (LLM) agents to interact with environments and solve multi-turn tasks. However, effective agentic RL remains challenging: sparse outcome-only rewards provide limited guidance for assigning credit to individual steps within long interaction trajectories. Existing approaches often introduce dense intermediate supervision, such as process reward models or auxiliary self-supervised signals, which increases supervision and tuning complexity and may limit generalization across tasks and domains. We present AEM, a supervision-free credit assignment method that adaptively modulates entropy dynamics during RL training to improve the exploration-exploitation trade-off. Since in agentic RL the environment is typically affected by a complete response, rather than an individual token, our analysis lifts entropy dynamics from the token level to the response level, aligning uncertainty estimation with the effective action granularity of LLM agents and reducing sensitivity to token-level sampling noise. We further show that entropy drift under natural-gradient updates is governed by the interaction between the sampled-response advantage and its relative surprisal. Motivated by this result, AEM derives a practical response-level uncertainty proxy and uses it to rescale advantages, leveraging the evolving balance between positive and negative samples to naturally transition from exploration to exploitation. Extensive experiments on ALFWorld, WebShop, and SWE-bench-Verified with models ranging from 1.5B to 32B demonstrate that AEM consistently improves strong RL baselines, including a +1.4\% gain when integrated into a state-of-the-art software-engineering RL training framework.

Document QA agents do not fail only at the final answer. They can also write unsupported claims into public working memory, and those writes can affect later retrieval and synthesis. We introduce ADMIT (Auditable Decoupled Memory-write admission for Intermediate Traces), a low-budget admission controller for public OCR/text evidence packs. ADMIT applies a deterministic substring-support check $S_{\text{str}}(\hat{y}, B) > 0$ to each public commit and allows one bounded RepairRead. If repair still fails, ADMIT uses a memory-fail-closed / output-fail-open rule: it writes a $\bot$ (NO_COMMIT) sentinel to memory while the final-answer surface may still fall back to the baseline answer. On held-out M3DocVQA, a retrained final-row router closes the answer-surface gap; ADMIT's distinguishing evidence is therefore memory-surface contamination, not final-answer accuracy. The lexical invariant guarantees zero-substring accepted intermediate contamination by construction; the contribution is therefore the public memory-write boundary as a low-cost control surface with decoupled fail actions, not the invariant itself. In a 7-arm dynamic stress diagnostic on three agents, ADMIT enforces the invariant and preserves raw EM at this operating point (zero contaminated trajectories on every agent and no detectable raw-EM loss relative to the fail-open ablation), whereas final-only baselines still leave $12$--$32%$ contamination. ADMIT enforces lexical auditability, not semantic truth: relation-level entailment, citation-aware faithfulness, and visual-region grounding are not claimed, and lower-density VisDoMSlide and text-only FRAMES are reported as boundary cases.

Large language models are increasingly used as scientific assistants, yet we still lack reliable ways to evaluate whether they can investigate a system rather than merely answer a prompt. A useful evaluation must let agents choose experiments, interpret finite evidence, and make causal decisions, while keeping the environment coherent across repeated observations and interventions and preserving objective ground truth. We introduce ACED-BENCH (Active Causal Evidence and Decision Benchmark), a neuro-symbolic benchmark built around this requirement. The key idea is to separate the readable scientific surface from the formal causal world: LLMs generate natural-language scenarios and task text, while hidden probabilistic graphical models define the causal graph, data-generating process, intervention semantics, validation checks, and oracle answers. This yields scientific worlds that are natural for agents to investigate but exact enough for scalable, programmatic evaluation. The main leaderboard ACED-CORE contains 180 tasks: 60 set-valued causal ancestry/descendancy questions and 120 constrained causal-decision tasks. We also report ACED-QUAL, a 150-question yes/no qualification check using the same hidden-PGM interface. We evaluate seven contemporary LLMs and two earlier-generation models across three active agent designs. ACED-QUAL checks whether agents can use the interface to draw conclusions from evidence, while ACED-CORE tests whether they can decide what evidence to collect and when to commit to an answer. Evidence-ledger analysis localizes failures within the investigation: agents test the wrong intervention, lose relevant evidence, or continue after sufficient evidence has appeared. These results suggest that reliable AI co-scientists need explicit support for experiment design, evidence retention, answer commitment, and stopping.

Foundation models are increasingly deployed as agentic systems that perceive context and act autonomously, raising the need to evaluate their moral behavior as active decision-makers. Existing benchmarks, however, largely ask models to judge human actions from third-person textual vignettes, leaving untested how models behave when they themselves become the acting agent. We introduce ABHBench, a multimodal benchmark built from \emph{Detroit: Become Human}, an interactive narrative game, to evaluate how foundation models make moral decisions in first-person human-android dilemmas. ABHBench places foundation models in branching narrative scenarios with visual observations, dialogue, memory, and action choices. Inspired by Asimov's Three Laws of Robotics, our framework measures Harmlessness, Obedience, and Self-preservation to capture the core tensions in human-AI relations, along with Autonomy, Stability, and Certainty to assess decision-making patterns. Across 12 foundation models, we find that moral behavior changes substantially under situated agentic context: 75\% of models shift significantly between a text-only abstraction and the matched multimodal DBH scenario, and reasoning-enhanced models show no monotonic gains in consistency or human-oriented behavior. These results highlight the necessity of evaluating moral behavior in situated agentic contexts rather than relying on third-person judgment alone.

Current embodied VLM evaluation relies on static, expert-defined, manually annotated benchmarks that exhibit severe redundancy and coverage imbalance. This labor‑intensive paradigm drains computational and annotation resources, inflates costs, and distorts model rankings, ultimately stifling iterative development. To address this, we propose Agentic Automatic Evaluation (A2Eval), the first agentic framework that automates benchmark curation and evaluation through two collaborative agents. The Data Agent autonomously induces capability dimensions and assembles a balanced, compact evaluation suite, while the Eval Agent synthesizes and validates executable evaluation pipelines, enabling fully autonomous, high-fidelity assessment. Evaluated across 10 benchmarks and 32 models, A2Eval compresses evaluation suites by 85\%, reduces overall computational costs by 77\%, and delivers up to a 4.6$\times$ speedup while preserving evaluation quality. Crucially, A2Eval corrects systematic ranking biases, improves human alignment to Spearman's $\rho=0.85$, and maintains high ranking fidelity (Spearman's $\rho=0.87$), establishing a new standard for high-fidelity, low-cost embodied assessment. Our code and data will be public upon acceptance.

Anchored fixed point and monotone equation methods, including Halpern iteration, extra anchored gradient, and their relatives, add a vanishing pull toward a reference point to obtain last-iterate guarantees. Existing anchored variants often achieve sharp last-iterate guarantees, but from the update-level perspective the placement of the anchor can be algorithm-specific and conceptually opaque. We show that anchoring admits a single operator-side construction: regularize the operator queried by the base method with a vanishing Tikhonov term, then run the unmodified base method. Applied to the Picard iteration, this recipe reproduces the Halpern iteration; applied to the forward step, extragradient, and Popov methods, it yields three variants whose anchor placements inherit the base method's query pattern. The four analyses share a residual recurrence, recovering the \(O(1/k)\) Halpern residual-norm convergence rate, giving \(O(1/\sqrt{k})\) for the regularized forward step, and giving \(O(1/k)\) for the regularized extragradient and Popov variants in the unconstrained monotone Lipschitz setting.

Material synthesis planning (MSP) remains a fundamental and underexplored bottleneck in AI-driven materials discovery, as it requires not only identifying suitable precursor materials but also designing coherent sequences of synthesis operations to realize a target material. Although several AI-based approaches have been proposed to address isolated subtasks of MSP, a methodology for solving the entire MSP task has yet to be established. We propose MSP-LLM, a structured LLM-based framework that formulates MSP as a two-stage process composed of two constituent subproblems: precursor prediction (PP) and synthesis operation prediction (SOP). Our approach introduces a discrete material class as an intermediate decision variable that organizes both tasks into a chemically consistent decision chain. For SOP, we further incorporate hierarchical precursor types as synthesis-relevant inductive biases and employ precursor constraint factorization that preserves precursor-related information in the autoregressive decoding state. Extensive experiments show that MSP-LLM consistently outperforms existing methods on both PP and SOP, as well as on the MSP task, demonstrating an effective and scalable framework with practical potential for autonomous MSP.