NeurIPS 2025 Papers

The molecular large language models have garnered widespread attention due to their promising potential on molecular applications. However, current molecular large language models face significant limitations in understanding molecules due to inadequate textual descriptions and suboptimal molecular representation strategies during pretraining. To address these challenges, we introduce KnowMol-100K, a large-scale dataset with 100K fine-grained molecular annotations across multiple levels, bridging the gap between molecules and textual descriptions. Additionally, we propose chemically-informative molecular representation, effectively addressing limitations in existing molecular representation strategies. Building upon these innovations, we develop KnowMol, a state-of-the-art multi-modal molecular large language model. Extensive experiments demonstrate that KnowMol achieves superior performance across molecular understanding and generation tasks.

In this paper, we reveal that most current efficient multimodal fine-tuning methods are hindered by a key limitation: they are directly borrowed from LLMs, often neglecting the intrinsic differences of multimodal scenarios and even affecting the full utilization of all modalities. Inspired by our empirical observation, we argue that unimodal adaptation and cross-modal adaptation are two essential parts for the effective fine-tuning of MLLMs. From this perspective, we propose Multimodal Low-rank Adaptation (MokA), a multimodal-aware efficient fine-tuning strategy that takes multimodal characteristics into consideration. It compresses unimodal information by modality-specific parameters while explicitly enhancing cross-modal interaction, ensuring both unimodal and cross-modal adaptation. Extensive experiments cover three representative multimodal scenarios (audio-visual-text, visual-text, and speech-text), and multiple LLM backbones (LLaMA2, Qwen2, Qwen2.5-VL, etc). Consistent improvements indicate the efficacy and versatility of the proposed method. Ablation studies and efficiency evaluation are also conducted to fully asses our method. Overall, we think MokA provides a more targeted solution for efficient adaptation of MLLMs, paving the way for further exploration.

Maritime object detection is essential for navigation safety, surveillance, and autonomous operations, yet constrained by two key challenges: the scarcity of annotated maritime data and poor generalization across various maritime attributes (e.g., object category, viewpoint, location, and imaging environment). To address these challenges, we propose Neptune-X, a data-centric generative-selection framework that enhances training effectiveness by leveraging synthetic data generation with task-aware sample selection. From the generation perspective, we develop X-to-Maritime, a multi-modality-conditioned generative model that synthesizes diverse and realistic maritime scenes. A key component is the Bidirectional Object-Water Attention module, which captures boundary interactions between objects and their aquatic surroundings to improve visual fidelity. To further improve downstream tasking performance, we propose Attribute-correlated Active Sampling, which dynamically selects synthetic samples based on their task relevance. To support robust benchmarking, we construct the Maritime Generation Dataset, the first dataset tailored for generative maritime learning, encompassing a wide range of semantic conditions. Extensive experiments demonstrate that our approach sets a new benchmark in maritime scene synthesis, significantly improving detection accuracy, particularly in challenging and previously underrepresented settings. The code is available at https://github.com/gy65896/Neptune-X.

Music-driven 3D dance generation has attracted increasing attention in recent years, with promising applications in choreography, virtual reality, and creative content creation. Previous research has generated promising realistic dance movement from audio signals. However, traditional methods underutilize genre conditioning, often treating it as auxiliary modifiers rather than core semantic drivers. This oversight compromises music-motion synchronization and disrupts dance genre continuity, particularly during complex rhythmic transitions, thereby leading to visually unsatisfactory effects. To address the challenge, we propose MEGADance, a novel architecture for music-driven 3D dance generation. By decoupling choreographic consistency into dance generality and genre specificity, MEGADance demonstrates significant dance quality and strong genre controllability. It consists of two stages: (1) High-Fidelity Dance Quantization Stage (HFDQ), which encodes dance motions into a latent representation by Finite Scalar Quantization (FSQ) and reconstructs them with kinematic-dynamic constraints, and (2) Genre-Aware Dance Generation Stage (GADG), which maps music into the latent representation by synergistic utilization of Mixture-of-Experts (MoE) mechanism with Mamba-Transformer hybrid backbone. Extensive experiments on the FineDance and AIST++ dataset demonstrate the state-of-the-art performance of MEGADance both qualitatively and quantitatively. Code will be released upon acceptance.

Recent advancements in text-to-video (T2V) diffusion models have enabled high-fidelity and realistic video synthesis. However, current T2V models often struggle to generate physically plausible content due to their limited inherent ability to accurately understand physics. We found that while the representations within T2V models possess some capacity for physics understanding, they lag significantly behind those from recent video self-supervised learning methods. To this end, we propose a novel framework called {VideoREPA}, which distills physics understanding capability from video understanding foundation models into T2V models by aligning token-level relations. This closes the physics understanding gap and enables more physics-plausible generation. Specifically, we introduce the {Token Relation Distillation (TRD) loss}, leveraging spatio-temporal alignment to provide soft guidance suitable for finetuning powerful pre-trained T2V models—a critical departure from prior representation alignment (REPA) methods. To our knowledge, VideoREPA is the first REPA method designed for finetuning T2V models and specifically for injecting physical knowledge. Empirical evaluations show that VideoREPA substantially enhances the physics commonsense of baseline method, CogVideoX, achieving significant improvement on relevant benchmarks and demonstrating a strong capacity for generating videos consistent with intuitive physics. Code and more video results are available at https://videorepa.github.io/.

Handwritten Mathematical Expression Recognition (HMER) remains a persistent challenge in Optical Character Recognition (OCR) due to the inherent freedom of symbol layouts and variability in handwriting styles. Prior methods have faced performance bottlenecks by proposing isolated architectural modifications, making them difficult to integrate coherently into a unified framework. Meanwhile, recent advances in pretrained vision-language models (VLMs) have demonstrated strong cross-task generalization, offering a promising foundation for developing unified solutions. In this paper, we introduce Uni-MuMER, which fully fine-tunes a VLM for the HMER task without modifying its architecture, effectively injecting domain-specific knowledge into a generalist framework. Our method integrates three data-driven tasks: Tree-Aware Chain-of-Thought (Tree-CoT) for structured spatial reasoning, Error-Driven Learning (EDL) for reducing confusion among visually similar characters, and Symbol Counting (SC) for improving recognition consistency in long expressions. Experiments on the CROHME and HME100K datasets show that Uni-MuMER achieves super state-of-the-art performance, outperforming the best lightweight specialized model SSAN by 16.31\% and the top-performing VLM Gemini2.5-flash by 24.42\% under zero-shot setting. Our datasets, models, and code are open-sourced at: https://github.com/BFlameSwift/Uni-MuMER

As AI models are deployed with increasing autonomy, it is important to ensure they do not take harmful actions unnoticed. As a potential mitigation, we investigate Chain-of-Thought (CoT) monitoring, wherein a weaker trusted monitor model continuously oversees the intermediate reasoning steps of a more powerful but untrusted model. We compare CoT monitoring to action-only monitoring, where only final outputs are reviewed, in a red-teaming setup where the untrusted model is instructed to pursue harmful side tasks while completing a coding problem. We find that while CoT monitoring is more effective than overseeing only model outputs in scenarios where action-only monitoring fails to reliably identify sabotage, reasoning traces can contain misleading rationalizations that deceive the CoT monitors, reducing performance in obvious sabotage cases. To address this, we introduce a hybrid protocol that independently scores model reasoning and actions, and combines them using a weighted average. Our hybrid monitor consistently outperforms both CoT and action-only monitors across all tested models and tasks, with detection rates twice higher than action-only monitoring for subtle deception scenarios.

Neuromorphic computing systems are set to revolutionize energy-constrained robotics by achieving orders-of-magnitude efficiency gains, while enabling native temporal processing. Spiking Neural Networks (SNNs) represent a promising algorithmic approach for these systems, yet their application to complex control tasks faces two critical challenges: (1) the non-differentiable nature of spiking neurons necessitates surrogate gradients with unclear optimization properties, and (2) the stateful dynamics of SNNs require training on sequences, which in reinforcement learning (RL) is hindered by limited sequence lengths during early training, preventing the network from bridging its warm-up period. We address these challenges by systematically analyzing surrogate gradient slope settings, showing that shallower slopes increase gradient magnitude in deeper layers but reduce alignment with true gradients. In supervised learning, we find no clear preference for fixed or scheduled slopes. The effect is much more pronounced in RL settings, where shallower slopes or scheduled slopes lead to a $\times2.1$ improvement in both training and final deployed performance. Next, we propose a novel training approach that leverages a privileged guiding policy to bootstrap the learning process, while still exploiting online environment interactions with the spiking policy. Combining our method with an adaptive slope schedule for a real-world drone position control task, we achieve an average return of 400 points, substantially outperforming prior techniques, including Behavioral Cloning and TD3BC, which achieve at most –200 points under the same conditions. This work advances both the theoretical understanding of surrogate gradient learning in SNNs and practical training methodologies for neuromorphic controllers demonstrated in real-world robotic systems.

Recent advancements in static feed-forward scene reconstruction have demonstrated significant progress in high-quality novel view synthesis. However, these models often struggle with generalizability across diverse environments and fail to effectively handle dynamic content. We present BTimer (short for Bullet Timer), the first motion-aware feed-forward model for real-time reconstruction and novel view synthesis of dynamic scenes. Our approach reconstructs the full scene in a 3D Gaussian Splatting representation at a given target (‘bullet’) timestamp by aggregating information from all the context frames. Such a formulation allows BTimer to gain scalability and generalization by leveraging both static and dynamic scene datasets. Given a casual monocular dynamic video, BTimer reconstructs a bullet-time scene within 150ms while reaching state-of-the-art performance on both static and dynamic scene datasets, even compared with optimization-based approaches.

Recent advancements in diffusion policies have demonstrated promising performance in decision-making tasks. To align these policies with human preferences, a common approach is incorporating Preference-based Reinforcement Learning (PbRL) into policy tuning. However, since preference data is practically collected from populations with different backgrounds, a key challenge lies in handling the inherent uncertainties in people's preferences during policy updates. To address this challenge, we propose the Diff-UAPA algorithm, designed for uncertainty-aware preference alignment in diffusion policies. Specifically, Diff-UAPA introduces a novel iterative preference alignment framework in which the diffusion policy adapts incrementally to preferences from different user groups. To accommodate this online learning paradigm, Diff-UAPA employs a maximum posterior objective, which aligns the diffusion policy with regret-based preferences under the guidance of an informative Beta prior. This approach enables direct optimization of the diffusion policy without specifying any reward functions, while effectively mitigating the influence of inconsistent preferences across different user groups. We conduct extensive experiments across both simulated and real-world robotics tasks, and diverse human preference configurations, demonstrating the robustness and reliability of Diff-UAPA in achieving effective preference alignment. The project page is available at https://anonymous.4open.science/w/Diff-UAPA-5827.

The increasing number of satellites and orbital debris has made space congestion a critical issue, threatening satellite safety and sustainability. Challenges such as collision avoidance, station-keeping, and orbital maneuvering require advanced techniques to handle dynamic uncertainties and multi-agent interactions. Reinforcement learning (RL) has shown promise in this domain, enabling adaptive, autonomous policies for space operations; however, many existing RL frameworks rely on custom-built environments developed from scratch, which often use simplified models and require significant time to implement and validate the orbital dynamics, limiting their ability to fully capture real-world complexities. To address this, we introduce OrbitZoo, a versatile multi-agent RL environment built on a high-fidelity industry standard library, that enables realistic data generation, supports scenarios like collision avoidance and cooperative maneuvers, and ensures robust and accurate orbital dynamics. The environment is validated against various real satellite constellations, including Starlink, achieving a Mean Absolute Percentage Error (MAPE) of 0.16% compared to real-world data. This validation ensures reliability for generating high-fidelity simulations and enabling autonomous and independent satellite operations.

The remarkable reasoning capability of large language models (LLMs) stems from cognitive behaviors that emerge through reinforcement with verifiable rewards. This work investigates how to transfer this principle to Multimodal LLMs (MLLMs) to unlock advanced visual reasoning. We introduce a two-stage paradigm built on Qwen2.5-VL-7B: a massive linguistic cold-start fine-tuning, followed by multimodal reinforcement learning (RL) spanning nearly 1,000 steps—surpassing all previous open-source efforts in scale. This pioneering work reveals three fundamental insights: 1) Behavior transfer emerges surprisingly early in cold start due to linguistic mental imagery. 2) Cold start broadly memorizes visual behaviors, while RL critically discerns and scales up effective patterns. 3) Transfer strategically favors high-utility behaviors such as visual reflection. Our resulting model, Open-Vision-Reasoner (OVR), achieves state-of-the-art performance on a suite of reasoning benchmarks, including 95.3% on MATH500, 51.8% on MathVision and 54.6% on MathVerse. We release our model, data, and training dynamics to catalyze the development of more capable, behavior-aligned multimodal reasoners.

Diffusion- and flow-based policies deliver state-of-the-art performance on long-horizon robotic manipulation and imitation-learning tasks. However, these controllers employ a fixed inference budget at every control step, regardless of task complexity, leading to computational inefficiency for simple subtasks while potentially underperforming on challenging ones. To address these issues, we introduce Difficulty-Aware Stochastic Interpolant Policy (DA-SIP), a framework that enables robotic controllers to adaptively adjust their integration horizon in real-time based on task difficulty. Our approach employs a difficulty classifier that analyzes RGB-D observations to dynamically select the step budget, the optimal solver variant, and ODE/SDE integration at each control cycle. DA-SIP builds upon the stochastic interpolant formulation to provide a unified framework that unlocks diverse training and inference configurations for diffusion- and flow-based policies. Through comprehensive benchmarks across diverse manipulation tasks, DA-SIP achieves 2.6-4.4× reduction in total computation time while maintaining task-success rates comparable to fixed maximum-computation baselines. By implementing adaptive computation within this framework, DA-SIP transforms generative robot controllers into efficient, task-aware systems that intelligently allocate inference resources where they provide the greatest benefit.

We describe KVLink, an approach for efficient key-value (KV) cache reuse in large language models (LLMs). In many LLM applications, different inputs can share overlapping context, such as the same retrieved document appearing in multiple queries. However, the LLMs still need to encode the entire context for each query, leading to redundant computation. In this paper, we investigate a new strategy to eliminate such inefficiency, where the KV cache of each document is precomputed independently. During inference, the KV caches of retrieved documents are concatenated, allowing the model to reuse cached representations instead of recomputing them. To mitigate the performance degradation when using KV caches computed independently for each document, KVLink introduces two key techniques: adjusting positional embeddings of the KV cache at inference to match the global position after concatenation, and using trainable special tokens to restore self-attention across independently encoded documents. Experiments across 7 datasets demonstrate that KVLink improves question answering accuracy by an average of 4% over state-of-the-art methods. Furthermore, by leveraging precomputed KV caches, our approach reduces time-to-first-token by up to 96% compared to standard LLM inference, making it a scalable and efficient solution for context reuse. Additionally, KVLink can be combined with KV cache compression to further save cache loading and storage overhead while outperforming the baselines.

As machine learning applications grow increasingly ubiquitous and complex, they face an increasing set of requirements beyond accuracy. The prevalent approach to handle this challenge is to aggregate a weighted combination of requirement violation penalties into the training objective. To be effective, this approach requires careful tuning of these hyperparameters (weights), involving trial-and-error and cross-validation, which becomes ineffective even for a moderate number of requirements. These issues are exacerbated when the requirements involve parities or equalities, as is the case in fairness and boundary value problems. An alternative technique uses constrained optimization to formulate these learning problems. Yet, existing approximation and generalization guarantees do not apply to problems involving equality constraints. In this work, we derive a generalization theory for equality-constrained statistical learning problems, showing that their solutions can be approximated using samples and rich parametrizations. Using these results, we propose a practical algorithm based on solving a sequence of *unconstrained*, *empirical* learning problems. We showcase its effectiveness and the new formulations enabled by equality constraints in fair learning, interpolating classifiers, and boundary value problems.

Pre-trained Large language models (LLMs) have shown impressive advancements in functional magnetic resonance imaging (fMRI) analysis and causal discovery. Considering the unique nature of the causal discovery field, which focuses on extracting causal graphs from observed data, research on LLMs in this field is still at an early exploratory stage. As a subfield of causal discovery, effective connectivity (EC) has received even less attention, and LLM-based approaches in EC remain unexplored. Existing LLM-based approaches for causal discovery typically rely on iterative querying to assess the causal influence between variable pairs, without any model adaptation or fine-tuning, making them ill-suited for handling the cross-modal gap and complex causal structures. To this end, we propose BrainEC-LLM, the first method to fine-tune LLMs for estimating brain EC from fMRI data. Specifically, multiscale decomposition mixing module decomposes fMRI time series data into short-term and long-term multiscale trends, then mixing them in bottom-up (fine to coarse) and top-down (coarse to fine) manner to extract multiscale temporal variations. And cross attention is applied with pre-trained word embeddings to ensure consistency between the fMRI input and pre-trained natural language. The experimental results on simulated and real resting-state fMRI datasets demonstrate that BrainEC-LLM can achieve superior performance when compared to state-of-the-art baselines.

Out-of-distribution (OOD) fine-tuning has emerged as a promising approach for anomaly segmentation. Current OOD fine-tuning strategies typically employ global-level objectives, aiming to guide segmentation models to accurately predict a large number of anomaly pixels. However, these strategies often perform poorly on small anomalies. To address this issue, we propose an instance-level OOD fine-tuning framework, dubbed LNOIB (Leaving No OOD Instance Behind). We start by theoretically analyzing why global-level objectives fail to segment small anomalies. Building on this analysis, we introduce a simple yet effective instance-level objective. Moreover, we propose a feature separation objective to explicitly constrain the representations of anomalies, which are prone to be smoothed by their in-distribution (ID) surroundings. LNOIB integrates these objectives to enhance the segmentation of small anomalies and serves as a paradigm adaptable to existing OOD fine-tuning strategies, without introducing additional inference cost. Experimental results show that integrating LNOIB into various OOD fine-tuning strategies yields significant improvements, particularly in component-level results, highlighting its strength in comprehensive anomaly segmentation.

We propose Flow-GRPO, the first method to integrate online policy gradient reinforcement learning (RL) into flow matching models. Our approach uses two key strategies: (1) an ODE-to-SDE conversion that transforms a deterministic Ordinary Differential Equation (ODE) into an equivalent Stochastic Differential Equation (SDE) that matches the original model's marginal distribution at all timesteps, enabling statistical sampling for RL exploration; and (2) a Denoising Reduction strategy that reduces training denoising steps while retaining the original number of inference steps, significantly improving sampling efficiency without sacrificing performance. Empirically, Flow-GRPO is effective across multiple text-to-image tasks. For compositional generation, RL-tuned SD3.5-M generates nearly perfect object counts, spatial relations, and fine-grained attributes, increasing GenEval accuracy from $63\%$ to $95\%$. In visual text rendering, accuracy improves from $59\%$ to $92\%$, greatly enhancing text generation. Flow-GRPO also achieves substantial gains in human preference alignment. Notably, very little reward hacking occurred, meaning rewards did not increase at the cost of appreciable image quality or diversity degradation.

For an LLM to correctly respond to an instruction it must understand both the semantics and the domain (i.e., subject area) of a given task-instruction pair. However, syntax can also convey implicit information. Recent work shows that \textit{syntactic templates}---frequent sequences of Part-of-Speech (PoS) tags---are prevalent in training data and often appear in model outputs. In this work we characterize syntactic templates, domain, and semantics in task-instruction pairs. We identify cases of spurious correlations between syntax and domain, where models learn to associate a domain with syntax during training; this can sometimes override prompt semantics. Using a synthetic training dataset, we find that the syntactic-domain correlation can lower performance (mean 0.51 +/- 0.06) on entity knowledge tasks in OLMo-2 models (1B-13B). We introduce an evaluation framework to detect this phenomenon in trained models, and show that it occurs on a subset of the FlanV2 dataset in open (OLMo-2-7B; Llama-4-Maverick), and closed (GPT-4o) models. Finally, we present a case study on the implications for LLM security, showing that unintended syntactic-domain correlations can be used to bypass refusals in OLMo-2-7B Instruct and GPT-4o. Our findings highlight two needs: (1) to explicitly test for syntactic-domain correlations, and (2) to ensure \textit{syntactic} diversity in training data, specifically within domains, to prevent such spurious correlations.