ICLR 2026 论文列表

Video understanding, including video captioning and retrieval, is still a great challenge for video-language models (VLMs). The existing video retrieval and caption benchmarks only include short descriptions, limits their ability of detailed video understanding evaluation. To address this problem, we present CaReBench, a testing benchmark for fine-grained video Captioning and Retrieval with 1,000 high-quality pairs of videos and human-annotated detailed captions. Uniquely, it provides manually separated spatial annotations and temporal annotations for each video. Based on this design, we introduce two evaluation metrics, ReBias and CapST, specifically tailored for video retrieval and video captioning tasks, respectively. These metrics enable a comprehensive investigation into the spatial and temporal biases inherent in VLMs. In addition, to handle both video retrieval and video captioning tasks in a unified framework, we develop a simple baseline based on a Multimodal Language Model (MLLM). By implementing a two-stage Supervised Fine-Tuning (SFT), we fully unlock the potential of MLLM, enabling it not only to generate detailed video descriptions but also to extract video features. Surprisingly, experimental results demonstrate that, compared to the CLIP-based models designed for retrieval and the popular MLLMs skilled in video captioning, our baseline shows competitive performance in both fine-grained video retrieval and video detailed captioning.

Considerable efforts have been devoted to machine learning (ML) for combinatorial optimization (CO) problems, especially on graphs. Compared to the active and well-established research for representation learning of text and vision, etc., it remains under-studied for the representation learning of CO problems, especially across different types. In this paper, we try to fill this gap (especially for NP-complete (NPC) problems, as they, in fact, can be reduced to one another). Our so-called ConRep4CO framework, performs contrastive learning by first transforming CO instances in various original forms into the form of Boolean satisfiability (SAT). This scheme is readily doable, especially for NPC problems, including those practical graph decision problems (GDPs) which are inherently related to their NP-hard optimization versions. Specifically, each positive pair of instances for contrasting consists of an instance in its original form and its corresponding transformed SAT form, while the negative samples are other instances not in correspondence. Extensive experiments on seven GDPs (most of which are NPC) show that ConRep4CO significantly improves the representation quality and generalizability to problem scale. Furthermore, we conduct extensive experiments on NP-hard optimization versions of the GDPs, including MVC, MIS, MC and MDS. The results show that introducing ConRep4CO can yield performance improvements of 61.27%, 32.20%, 36.46%, and 45.29% in objective value gaps compared to problem-specific baselines, highlighting the potential of ConRep4CO as a unified pre-training paradigm for CO problems.

Uncertainty Quantification (UQ) research has primarily focused on closed-book factual question answering (QA), while contextual QA remains unexplored, despite its importance in real-world applications. In this work, we focus on UQ for the contextual QA task and propose a theoretically grounded approach to quantify \emph{epistemic uncertainty}. We begin by introducing a task-agnostic, token-level uncertainty measure defined as the cross-entropy between the predictive distribution of the given model and the unknown true distribution. By decomposing this measure, we isolate the epistemic component and approximate the true distribution by a perfectly prompted, idealized model. We then derive an upper bound for epistemic uncertainty and show that it can be interpreted as semantic feature gaps in the given model’s hidden representations relative to the ideal model. We further apply this generic framework to the contextual QA task and hypothesize that three features approximate this gap: \emph{context-reliance} (using the provided context rather than parametric knowledge), \emph{context comprehension} (extracting relevant information from context), and \emph{honesty} (avoiding intentional lies). Using a top-down interpretability approach, we extract these features by using only a small number of labeled samples and ensemble them to form a robust uncertainty score. Experiments on multiple QA benchmarks in both in-distribution and out-of-distribution settings show that our method substantially outperforms state-of-the-art unsupervised (sampling-free and sampling-based) and supervised UQ methods, achieving up to a 13-point PRR improvement while incurring a negligible inference overhead. The code is available at https://github.com/Ybakman/Feature-Gaps.

While model-based reinforcement learning (MBRL) improves sample efficiency by learning world models from raw observations, existing methods struggle to generalize across structurally similar scenes and remain vulnerable to spurious variations such as textures or color shifts. From a cognitive science perspective, humans segment continuous sensory streams into discrete events and rely on these key events for decision-making. Motivated by this principle, we propose the Event-Aware World Model (EAWM), a general framework that learns event-aware representations to streamline policy learning without requiring handcrafted labels. EAWM employs an automated event generator to derive events from raw observations and introduces a Generic Event Segmentor (GES) to identify event boundaries, which mark the start and end time of event segments. Through event prediction, the representation space is shaped to capture meaningful spatio-temporal transitions. Beyond this, we present a unified formulation of seemingly distinct world model architectures and show the broad applicability of our methods. Experiments on Atari 100K, Craftax 1M, and DeepMind Control 500K, DMC-GB2 500K demonstrate that EAWM consistently boosts the performance of strong MBRL baselines by 10\%–45\%, setting new state-of-the-art results across benchmarks. Our code is released at [https://github.com/MarquisDarwin/EAWM](https://github.com/MarquisDarwin/EAWM).

We introduce the first best-of-both-worlds algorithm for contextual combinatorial semi-bandits that simultaneously guarantees $\widetilde{\mathcal{O}}(\sqrt{T})$ regret in the adversarial regime and $\widetilde{\mathcal{O}}(\ln T)$ regret in the corrupted stochastic regime. Our approach builds on the Follow-the-Regularized-Leader (FTRL) framework equipped with a Shannon entropy regularizer, yielding a flexible method that admits efficient implementations. Beyond regret bounds, we tackle the practical bottleneck in FTRL (or, equivalently, Online Stochastic Mirror Descent) arising from the high-dimensional projection step encountered in each round of interaction. By leveraging the Karush-Kuhn-Tucker conditions, we transform the $K$-dimensional convex projection problem into a single-variable root-finding problem, dramatically accelerating each round. Empirical evaluations demonstrate that this combined strategy not only attains the attractive regret bounds of best-of-both-worlds algorithms but also delivers substantial per-round speed-ups, making it well-suited for large-scale, real-time applications.

Multimodal large language models (MLLMs) have achieved strong performance on vision-language tasks, yet often suffer from inefficiencies due to redundant visual tokens. Existing token merging methods reduce sequence length but frequently disrupt spatial layouts and temporal continuity by disregarding positional relationships. In this work, we propose a novel encoding operator dubbed as **P**ositional **P**reservation **E**mbedding (**PPE**), which has the main hallmark of preservation of spatiotemporal structure during visual token compression. PPE explicitly introduces the disentangled encoding of 3D positions in the token dimension, enabling each compressed token to encapsulate different positions from multiple original tokens. Furthermore, we show that PPE can effectively support cascade clustering --- a progressive token compression strategy that leads to better performance retention. PPE is a parameter-free and generic operator that can be seamlessly integrated into existing token merging methods without any adjustments. Applied to state-of-the-art token merging framework, PPE achieves consistent improvements of 2\%~5\% across multiple vision-language benchmarks, including MMBench (general vision understanding), TextVQA (layout understanding) and VideoMME (temporal understanding). These results demonstrate that preserving positional cues is critical for efficient and effective MLLM reasoning. Our code is available at https://github.com/MouxiaoHuang/PPE

Training large language models (LLMs) for complex reasoning via Reinforcement Learning with Verifiable Rewards (RLVR) is effective but limited by reliance on costly, domain-specific supervision. We explore Reinforcement Learning from Internal Feedback (RLIF), a framework that enables LLMs to learn from intrinsic signals without external rewards or labeled data. We propose Intuitor, an RLIF method that uses a model's own confidence—termed self-certainty—as its sole reward signal. Intuitor replaces external rewards in Group Relative Policy Optimization (GRPO) with self-certainty scores, enabling fully unsupervised learning. Experiments demonstrate that Intuitor matches GRPO's performance on mathematical benchmarks and achieves competitive, sometimes better, generalization to out-of-domain tasks like code generation, without requiring gold solutions or test cases. Our findings show that intrinsic model signals can drive effective learning across domains, offering a scalable alternative to RLVR for autonomous AI systems where verifiable rewards are unavailable. Code is available in the supplementary materials.

The evaluation of discourse-level translation in expert domains remains inadequate, despite its centrality to knowledge dissemination and cross-lingual scholarly communication. While these translations demand discourse-level coherence and strict terminological precision, current evaluation methods predominantly focus on segment-level accuracy and fluency. To address this limitation, we introduce DiscoX, a new benchmark for discourse-level and expert-level Chinese-English translation. It comprises 200 professionally-curated texts from 7 domains, with an average length exceeding 1700 tokens. To evaluate performance on DiscoX, we also develop Metric-S, a reference-free system that provides fine-grained automatic assessments across accuracy, fluency, and appropriateness. Metric-S demonstrates strong consistency with human judgments, significantly outperforming existing metrics. Our experiments reveal a remarkable performance gap: even the most advanced LLMs still trail human experts on these tasks. This finding validates the difficulty of DiscoX and underscores the challenges that remain in achieving professional-grade machine translation. The proposed benchmark and evaluation system provide a robust framework for more rigorous evaluation, facilitating future advancements in LLM-based translation. Our data and code are available at https://anonymous.4open.science/r/DiscoX-5F18.

Due to practical constraints such as partial observability and limited communication, Centralized Training with Decentralized Execution (CTDE) has become the dominant paradigm in cooperative Multi-Agent Reinforcement Learning (MARL). However, existing CTDE methods often underutilize centralized training or lack theoretical guarantees. We propose Multi-Agent Guided Policy Optimization (MAGPO), a novel framework that better leverages centralized training by integrating centralized guidance with decentralized execution. MAGPO uses an autoregressive joint policy for scalable, coordinated exploration and explicitly aligns it with decentralized policies to ensure deployability under partial observability. We provide theoretical guarantees of monotonic policy improvement and empirically evaluate MAGPO on 43 tasks across 6 diverse environments. Results show that MAGPO consistently outperforms strong CTDE baselines and matches or surpasses fully centralized approaches, offering a principled and practical solution for decentralized multi-agent learning.

Computational replication of Chinese calligraphy, a cornerstone of cultural heritage, remains challenging. Existing methods split into two flawed camps: some render high-quality isolated characters yet miss page-level aesthetics (ligatures, spacing, scale), while others attempt page/column synthesis but sacrifice calligraphic correctness. We introduce UniCalli, a unified diffusion framework for column-level recognition and generation. Training both tasks in one model is deliberate: recognition constrains the generator to preserve character identity and stroke structure, while generation supplies strong style/layout priors—together fostering concept-level abstractions (radicals, stroke configurations) that improve both tasks under long-tail, limited-label regimes. We curate a dataset of 8,000+ digitized pieces, with ~4,000 densely annotated (script labels, character boxes, transcriptions). UniCalli employs asymmetric noising and a rasterized box map to inject spatial priors, and is trained on a mix of synthetic, labeled, and unlabeled data. The model is robust to rare styles, better disentangles style from script, and attains state-of-the-art generative quality with clear gains in ligature continuity and layout fidelity, alongside stronger recognition. The framework extends to other ancient scripts, demonstrated by successful transfer to Oracle bone inscriptions and Egyptian hieroglyphs. Code and data will be released.

Existing language model benchmarks provide contradictory model rankings, even for benchmarks that aim to capture similar skills. This dilemma of conflicting rankings hampers model selection, clouds model comparisons, and adds confusion to a growing ecosystem of competing models. In this paper, we take a different perspective on model comparison: instead of relying on out-of-the-box performance via direct evaluation, we compare model potential by providing each model with identical benchmark-specific fine-tuning before evaluation. We call this approach train-before-test. Our primary contribution is a comprehensive empirical evaluation of model potential across 24 benchmarks and 61 models. First, we demonstrate that model potential rankings obtained through train-before-test exhibit remarkable consistency across all benchmarks. Whereas traditional rankings demonstrate little external validity under direct evaluation, they enjoy a significant degree of external validity when applying train-before-test: model potential rankings transfer gracefully from one benchmark to another. Second, train-before-test restores the connection between perplexity and downstream task performance, lost under direct evaluation. Remarkably, even pre-finetuning perplexity of a base model predicts post-finetuning downstream performance, suggesting that ranking consistency reflects inherent model potential rather than fine-tuning artifacts. Finally, train-before-test reduces the model-score matrix to essentially rank one, indicating that model potential is dominated by one latent factor, uncovered by train-before-test. While direct evaluation remains useful for assessing deployment-ready performance, train-before-test provides a complementary lens for understanding achievable performance of models after adaptation.

The pre-training of large language models (LLMs) relies on massive text datasets sourced from diverse and difficult-to-curate origins. Although membership inference attacks and hidden canaries have been explored to trace data usage, such methods rely on *regurgitation* of training data, which LM providers try to limit. In this work, we demonstrate that *indirect data poisoning* (where the targeted behavior is absent from training data) is not only feasible against LLMs but also allows to effectively protect a dataset and trace its use. Using gradient-based optimization prompt-tuning, we craft poisons to make a model learn arbitrary *secret sequences*: secret responses to secret prompts that are **absent from the training corpus**.\ We validate our approach on language models pre-trained from scratch and show that less than 0.005\% of poisoned tokens are sufficient to covertly make a LM learn a *secret* and detect it with extremely high confidence ( $p < 10^{-55}$ ) with a theoretically certifiable scheme. Crucially, this occurs without performance degradation (on LM benchmarks) and despite secrets **never appearing in the training set**.

While current Multimodal Large Language Models (MLLMs) have demonstrated proficiency in reasoning tasks such as mathematics and logic, their capacity for long-chain reflective reasoning, a prerequisite for solving complex real-world problems, remains largely underexplored. In this work, we first conduct an extensive empirical investigation to evaluate this capability. Leveraging a carefully designed data synthesis engine, we construct MM-HELIX, a multimodal benchmark consisting 1260 samples of 42 challenging synthetic tasks that require iterative thinking and backtracking. Empirical results on this benchmark reveal that existing MLLMs exhibit significant performance deficits in long-chain reflective reasoning. To address this limitation, we generate post-training data and further explore learning paradigms for exploiting such data. We first develop the Step-Elicited Response Generation pipeline to create MM-HELIX-100K, a large-scale dataset of 100k high-quality, reflective reasoning traces for instruction-tuning stage. Given that standard Reinforcement Learning fails on complex tasks due to sparse reward signals and catastrophic forgetting after Supervised Fine-Tuning, we propose Adaptive Hybrid Policy Optimization (AHPO), a novel training strategy that dynamically unifies offline supervision and online optimization into a single stage. This strategy enables the model to learn from expert data when rewards are sparse and conduct independent exploration once proficient. When applied to the Qwen2.5-VL-7B baseline, our method achieves a +18.6\% accuracy improvement on MM-HELIX benchmark and demonstrates strong generalization with a +5.7\% average performance gain on general mathematic and logic tasks. Our work demonstrate that reflective reasoning in MLLMs can be effectively learned and generalized, paving the way for developing more capable MLLMs.

Text-to-video diffusion models have enabled high-quality video synthesis, yet often fail to generate temporally coherent and physically plausible motion. A key reason is the models' insufficient understanding of complex motions that natural videos often entail. Recent works tackle this problem by aligning diffusion model features with those from pretrained video encoders. However, these encoders mix video appearance and dynamics into entangled features, limiting the benefit of such alignment. In this paper, we propose a motion-centric alignment framework that learns a disentangled motion subspace from a pretrained video encoder. This subspace is optimized to predict ground-truth optical flow, ensuring it captures true motion dynamics. We then align the latent features of a text-to-video diffusion model to this new subspace, enabling the generative model to internalize motion knowledge and generate more plausible videos. Our method improves the physical commonsense in a state-of-the-art video diffusion model, while preserving adherence to textual prompts, as evidenced by empirical evaluations on VideoPhy, VideoPhy2, VBench, and VBench-2.0, along with a user study.

We study the problem of assigning operations in a dataflow graph to devices to minimize execution time in a work-conserving system, with emphasis on complex machine learning workloads. Prior learning-based approaches face three limitations: (1) reliance on bulk-synchronous frameworks that under-utilize devices, (2) learning a single placement policy without modeling the system dynamics, and (3) depending solely on reinforcement learning in pre-training while ignoring optimization during deployment. We propose Doppler, a three-stage framework with two policies—SEL for selecting operations and PLC for placing them on devices. Doppler consistently outperforms baselines by reducing execution time and improving sampling efficiency through faster per-episode training. Our results show that DOPPLER achieves up to 52.7% lower execution times than the best baseline.

Deep neural networks (DNNs) are highly vulnerable to adversarial attacks, where small, carefully crafted perturbations are added to input images to cause misclassification. These perturbations are particularly effective when concentrated in sensitive regions of an image that strongly influence the model’s prediction. However, in decision-based black-box settings, where only the top-1 predicted label is observable and query budgets are strictly limited, identifying sensitive regions becomes extremely challenging. This issue is critical because without accurate region information, decision-based attacks cannot refine adversarial examples effectively, limiting both their efficiency and accuracy. We propose Sensitive Region Identification, SeRI, the first decision-based method that assigns a continuous sensitivity score to each image pixel. It enables fine-grained region discovery and substantially improves the efficiency of adversarial attacks, all without access to gradients, confidence scores, or surrogate models. SeRI progressively partitions the image into finer sub-regions and refines a continuous sensitivity score to capture their true importance. At each iteration, it generates two perturbation variants of the selected region by scaling its magnitude up or down, and compares their decision boundaries to derive an accurate, continuous characterization of pixel sensitivity. SeRI further divides selected region into smaller sub-regions, recursively refining the search for sensitive areas. This recursive refinement process enables more precise sensitivity estimation through fine-grained analysis, distinguishing SeRI from prior binary or one-shot region selection approaches. Experiments on two benchmark datasets show that SeRI significantly enhances state-of-the-art decision-based attacks in both targeted and non-targeted attack scenarios. Additionally, SeRI generates precise heatmaps that identify sensitive image regions. The code is available at https://anonymous.4open.science/r/SeRI-5310.

Large Language Models (LLMs) have recently improved mathematical reasoning through Reinforcement Learning with Verifiable Reward (RLVR). However, existing RLVR algorithms require large query budgets, making annotation costly. We investigate whether fewer but more informative queries can yield similar or superior performance, introducing active learning (AL) into RLVR. We identify that classic AL sampling strategies fail to outperform random selection in this setting, due to ignoring \textbf{objective uncertainty} when only selecting by subjective uncertainty. This work proposes an \textbf{uncertainty consistency} metric to evaluate how well subjective uncertainty aligns with objective uncertainty. In the offline setting, this alignment is measured using the Point-Biserial Correlation Coefficient (PBC). For online training, because of limited sampling and dynamically shifting output distributions, PBC estimation is difficult. Therefore, we introduce a new online variant, computed from normalized advantage and subjective uncertainty. Theoretically, we prove that the online variant is strictly negatively correlated with offline PBC and supports better sample selection. Experiments show our method consistently outperforms random and classic AL baselines, achieving full-dataset performance while training on only 30\% of the data, effectively reducing the cost of RLVR for reasoning tasks.\footnote{The code is available at \hyperref[https://github.com/yihao-123/uncertainty-consistency]{https://github.com/yihao-123/uncertainty-consistency}. }

Weight-space models learn directly from the parameters of neural networks, enabling tasks such as predicting their accuracy on new datasets. Naive methods -- like applying MLPs to flattened parameters -- perform poorly, making the design of better weight-space architectures a central challenge. While prior work leveraged permutation symmetries in standard networks to guide such designs, no analogous analysis or tailored architecture yet exists for Kolmogorov–Arnold Networks (KANs). In this work, we show that KANs share the same permutation symmetries as MLPs, and propose the KAN-graph, a graph representation of their computation. Building on this, we develop WS-KAN, the first weight-space architecture that learns on KANs, which naturally accounts for their symmetry. We analyze WS-KAN’s expressive power, showing it can replicate an input KAN’s forward pass - a standard approach for assessing expressiveness in weight-space architectures. We construct a comprehensive ``zoo'' of trained KANs spanning diverse tasks, which we use as benchmarks to empirically evaluate WS-KAN. Across all tasks, WS-KAN consistently outperforms structure-agnostic baselines, often by a substantial margin.

Chain-of-Thought (CoT) reasoning often fails to faithfully reflect a language model's underlying decision process. We address this by introducing a \emph{Markovian} language model framework with an autoencoder-style \emph{reasoning bottleneck}: it creates a text-based bottleneck where CoT serves as an intermediate representation, forcing the model to compress essential reasoning into interpretable text before making predictions, in the sense of learning short intermediate descriptions that make answers easy to compute from questions. We train this system with a GRPO-style policy gradient algorithm using parallel sampling, a frozen baseline CoT$'$, within-batch standardized advantages, and actor-reward (chain-rule) gradients. On QA tasks, Markovian training recovers most of the gains of a non-Markovian GRPO variant while forcing the model to answer from the CoT alone (e.g., GSM8K: 19.6\% $\to$ 57.1\%; ARC-Challenge: 36.1\% $\to$ 79.9\%; on average only $\approx$3-4 pp below a non-Markovian upper bound). Perturbation analyses across types and severities show that Markovian models incur systematically larger log-probability drops under CoT corruption than matched Non-Markovian baselines, indicating stronger causal reliance on the CoT. Cross-model evaluation confirms that learned CoTs generalize across architectures, suggesting they capture transferable reasoning patterns rather than model-specific artifacts.