ICLR 2026 论文列表

Discrete diffusion models have demonstrated great promise in modeling various sequence data, ranging from human language to biological sequences. Inspired by the success of RL in language models, there is growing interest in further improving the models by alignment with a certain reward. In this work, we propose an offline preference optimization method to approach trajectory alignment for discrete diffusion models. Instead of applying the reward on the final output and backpropagating the gradient to the entire denoising process, we decompose the problem into a set of stepwise alignment objectives by matching the per-step posterior. This framework enables efficient diffusion optimization, is compatible with arbitrary reward functions, and importantly, yields an equivalent optimal solution under additive factorization of the trajectory reward. Experiments across multiple domains including DNA sequence design, protein inverse folding, and language modeling consistently demonstrate the superiority of our approach. Notably, it achieves an up to 12\% improvement over the most competitive RL-based baseline in terms of predicted activity on DNA sequence design, and further improves the GSM8K score from 78.6 to 81.2 on LLaDA-8B-Instruct for language modeling.

Modeling event sequences with multiple event types with marked temporal point processes (MTPPs) provides a principled way to uncover governing rules and predict future events. Current neural network approaches to MTPP inference rely on training separate, specialized models for each target system. We pursue a radically different approach: drawing on amortized inference and in-context learning, we pretrain a deep neural network to infer, *in-context*, the conditional intensity functions of event histories from a context defined by sets of event sequences. Pretraining is performed on a large synthetic dataset of MTPPs sampled from a broad distribution over Hawkes processes. Once pretrained, our Foundation Inference Model for Point Processes (FIM-PP) can estimate MTPPs from real-world data without any additional training, or be rapidly finetuned to target systems. Experiments show that this amortized approach matches the performance of specialized models on next-event prediction across common benchmark datasets. We provide the pretrained model weights with the supplementary material.

Model-level explanations for Graph Neural Networks (GNNs) aim to identify class-discriminative motifs that capture how a classifier recognizes a target class. Because the true motifs relied on by the classifier are unobservable, most approaches evaluate explanations by their target class score. However, class score alone is not sufficient as high-scoring explanations may be pathological or may fail to reflect the full range of motifs recognized by the classifier. To bridge this gap, this work introduces sufficiency risk as a formal criterion for whether explanations adequately represent the classifier’s reasoning, and derives distribution-free certificates that upper-bound this risk. Building on this foundation, three metrics are introduced: Coverage, Greedy Gain Area (GGA), and Overlap which operationalize the certificates to assess sufficiency, efficiency, and redundancy in explanations. To ensure practical utility, finite-sample concentration bounds are developed for these metrics, providing confidence intervals that enable statistically reliable comparison between explainers. Experiments on synthetic data and with three state-of-the-art explainers on four real-world datasets demonstrate that these metrics reveal differences in explanation quality hidden by class scores alone. Designed to complement class score, they constitute the first theoretically certified framework for evaluating model-level explanations of GNNs.

Modern deep neural networks remain challenging to interpret due to the opacity of their latent representations, impeding model understanding, debugging, and debiasing. Concept Embedding Models (CEMs) address this by mapping inputs to human-interpretable concept representations from which tasks can be predicted. Yet, CEMs fail to represent inter-concept relationships and require concept annotations at different granularities during training, limiting their applicability. In this paper, we introduce *Hierarchical Concept Embedding Models* (HiCEMs), a new family of CEMs that explicitly model concept relationships through hierarchical structures. To enable HiCEMs in real-world settings, we propose *Concept Splitting*, a method for automatically discovering finer-grained sub-concepts from a pretrained CEM’s embedding space without requiring additional annotations. This allows HiCEMs to generate fine-grained explanations from limited concept labels, reducing annotation burdens. Our evaluation across multiple datasets, including a user study and experiments on *PseudoKitchens*, a newly proposed concept-based dataset of 3D kitchen renders, demonstrates that (1) Concept Splitting discovers human-interpretable sub-concepts absent during training that can be used to train highly accurate HiCEMs, and (2) HiCEMs enable powerful test-time concept interventions at different granularities, leading to improved task accuracy.

Generative models for speech synthesis face a fundamental trade-off: discrete tokens ensure stability but sacrifice expressivity, while continuous signals retain acoustic richness but suffer from error accumulation due to task entanglement. This challenge has driven the field towards multi-stage pipelines that rely on pre-trained discrete speech tokenizers, but these create a semantic-acoustic divide, limiting holistic and expressive speech generation. We resolve these dilemma through hierarchical semantic-acoustic modeling with semi-discrete residual representations.Our framework introduces a differentiable quantization bottleneck that induces natural specialization: a Text-Semantic Language Model (TSLM) generates semantic-prosodic plans, while a Residual Acoustic Model (RALM) recovers fine-grained acoustic details.This hierarchical semantic-acoustic representation guides a local diffusion-based decoder to generate high-fidelity speech latents. Critically, the entire architecture is trained end-to-end under a simple diffusion objective, eliminating dependency on external discrete speech tokenizers. Trained on over 1 million hours of speech, our 0.5B-parameter model achieves state-of-the-art zero-shot TTS performance among open-source systems, demonstrating that our approach delivers expressive and stable synthesis. Audio samples are available at: https://voxcpm.github.io/VoxCPM-demopage/.

Recent studies on end-to-end (E2E) speech generation with large language models (LLMs) have attracted significant community attention, with multiple works extending text-based LLMs to generate discrete speech tokens. Existing E2E approaches primarily fall into two categories: (1) Methods that generate discrete speech tokens independently without incorporating them into the LLM’s autoregressive process, resulting in text generation being unaware of concurrent speech synthesis. (2) Models that generate interleaved or parallel speech-text tokens through joint autoregressive modeling, enabling mutual modality awareness during generation. This paper presents DrVoice, a parallel speech-text voice conversation model based on joint autoregressive modeling, featuring dual-resolution speech representations. Notably, while current methods utilize mainly 12.5Hz input audio representation, our proposed dual-resolution mechanism reduces the input frequency for the LLM to 5Hz, significantly reducing computational cost and alleviating the frequency discrepancy between speech and text tokens and in turn better exploiting LLMs’ capabilities. Experimental results demonstrate that DrVoice-7B establishes new state-of-the-art (SOTA) on prominent speech benchmarks including OpenAudioBench, VoiceBench, UltraEval-Audio and Big Bench Audio, making it a leading open-source speech foundation model in ∼7B models.

Systematic generalization refers to the capacity to understand and generate novel combinations from known components. Despite recent progress by large language models (LLMs) across various domains, these models often fail to extend their knowledge to novel compositional scenarios, revealing notable limitations in systematic generalization. There has been an ongoing debate about whether neural networks possess the capacity for systematic generalization, with recent studies suggesting that meta-learning approaches designed for compositionality can significantly enhance this ability. However, these insights have largely been confined to linguistic problems, leaving their applicability to other tasks an open question. In this study, we extend meta-learning for compositionality to the domain of abstract spatial reasoning. To this end, we introduce $\textit{Compositional-ARC}\textemdash{}$a dataset designed to evaluate the capacity of models to systematically generalize from known geometric transformations (e.g., translation, rotation) of abstract two-dimensional objects to novel combinations of these transformations (e.g., translation+rotation). Our results show that a small transformer-based encoder-decoder model, trained via meta-learning for compositionality, can systematically generalize to previously unseen transformation compositions. Notably, despite having only 5.7M parameters, this model significantly outperforms state-of-the-art LLMs$\textemdash{}$including o3-mini, GPT-4o, and Gemini 2.0 Flash, which fail to exhibit similar systematic behavior$\textemdash{}$and performs on par with the winning model of the ARC prize 2024, an 8B-parameter LLM trained via test-time training. Our findings highlight the effectiveness of meta-learning in promoting systematicity beyond linguistic tasks, suggesting a promising direction toward more robust and generalizable models.

Large language models and vision-language models score high on many medical QA benchmarks; however, real-world clinical reasoning remains challenging because cases often involve multiple images and require cross-view fusion. We present MedThinkVQA, a benchmark that asks models to think with multiple images: read each image, merge evidence across views, and pick a diagnosis with stepwise supervision. We make three parts explicit: multi-image questions, expert-annotated stepwise supervision, and beyond-accuracy evaluation. Only MedThinkVQA combines all these parts in one expert-annotated benchmark. The dataset has 8,481 cases in total, with 751 test cases, and on average 6.51 images per case; it is expert-annotated and, at this level, larger and more image-dense than prior work (earlier maxima < 1.43 images per case). On the test set, GPT-5 achieves 57.39% accuracy, approximately 15 percentage points below the strongest result on the most challenging prior benchmark of a similar kind, while other strong models are lower (Qwen2.5-VL-32B: 39.54%, MedGemma-27B: 37.55%, InternVL3.5-38B: 43.14%). Giving expert findings and summaries brings clear gains, but using models' self-generated ones brings small or negative gains. Step-level evaluation shows where models stumble: errors center on image reading and cross-view integration in both decisive and non-decisive steps (>70%); when a step is decisive for the final choice, reasoning slips become more common (32.26%), while scenario and pure-knowledge slips are relatively rare (<10%). These patterns isolate and quantify the core obstacle: extracting and integrating cross-image evidence, rather than language-only inference.

Collaborative perception integrates multi-agent perspectives to enhance the sensing range and overcome occlusion issues. While existing multimodal approaches leverage complementary sensors to improve performance, they are highly prone to failure—especially when a key sensor like LiDAR is unavailable. The root cause is that feature fusion leads to semantic mismatches between single-modality features and the downstream modules. This paper addresses this challenge for the first time in the field of collaborative perception, introducing **Si**ngle-**M**odality-**O**perable Multimodal Collaborative Perception (**SiMO**). By adopting the proposed **L**ength-**A**daptive **M**ulti-**M**od**a**l Fusion (**LAMMA**), SiMO can adaptively handle remaining modal features during modal failures while maintaining consistency of the semantic space. Additionally, leveraging the innovative "Pretrain-Align-Fuse-RD" training strategy, SiMO addresses the issue of modality competition—generally overlooked by existing methods—ensuring the independence of each individual modality branch. Experiments demonstrate that SiMO effectively aligns multimodal features while simultaneously preserving modality-specific features, enabling it to maintain optimal performance across all individual modalities.

On-device inference offers privacy, offline use, and instant response, but consumer hardware restricts large language models (LLMs) to low throughput and capability. To overcome this challenge, we present prima.cpp, a distributed on-device inference system that runs 30-70B LLMs on consumer home clusters with mixed CPUs/GPUs, insufficient RAM/VRAM, slow disks, Wi-Fi links, and heterogeneous OSs. We introduce pipelined-ring parallelism (PRP) to overlap disk I/O with compute and communication, and address the prefetch-release conflict in mmap-based offloading. We further propose Halda, a heterogeneity-aware scheduler that co-optimizes per-device CPU/GPU workloads and device selection under RAM/VRAM constraints. On four consumer home devices, a 70B model reaches 674 ms/token TPOT with <6% memory pressure, and a 32B model with speculative decoding achieves 26 tokens/s. Compared with llama.cpp, exo, and dllama, our proposed prima.cpp achieves 5-17× lower TPOT, supports fine-grained model sizes from 8B to 70B, ensures broader cross-OS and quantization compatibility, and remains OOM-free, while also being Wi-Fi tolerant, privacy-preserving, and hardware-independent. The code is available at https://gitee.com/zonghang-li/prima.cpp.

Masked discrete diffusion models have been gaining popularity recently, and classifier-free guidance, just like its continuous counterpart, has been proposed to enable efficacious conditional generation by discrete diffusion. To quantify the precise effect of discrete guidance, this article considers masked discrete diffusion with arbitrary data distribution in low dimension, so that the distribution that guided masked discrete diffusion samples from, as well as the sampling dynamics, can be analytically and exactly quantified and interpreted. When the full data distribution is a mixture over classes and the goal is to sample from a specific class, guidance amplifies class-specific regions while suppresses regions shared with other classes. This effect depends on the guidance strength $w$ and induces distinct covariance structures in the sampled distribution. Notably, we observe quantitatively different behaviors in $1$D and $2$D. We also show that for large $w$, the decay rate of the total variation ($\text{TV}$) along the reverse dynamics is double-exponential in $w$ for both $1$D and $2$D. These findings highlight the role of guidance, not just in shaping the output distribution, but also in controlling the dynamics of the sampling trajectory. Our theoretical analysis is supported by experiments that illustrate the geometric effects of guidance and its impact on convergence.

Diffusion models have recently emerged as powerful tools for camera simulation, enabling both geometric transformations and realistic optical effects. Among these, image-based bokeh rendering has shown promising results, but diffusion for video bokeh remains unexplored. Existing image-based methods are plagued by temporal flickering and inconsistent blur transitions, while current video editing methods lack explicit control over the focus plane and bokeh intensity. These issues limit their applicability for controllable video bokeh. In this work, we propose a one-step diffusion framework for generating temporally coherent, depth-aware video bokeh rendering. The framework employs a multi-plane image (MPI) representation adapted to the focal plane to condition the video diffusion model, thereby enabling it to exploit strong 3D priors from pretrained backbones. To further enhance temporal stability, depth robustness, and detail preservation, we introduce a progressive training strategy. Experiments on synthetic and real-world benchmarks demonstrate superior temporal coherence, spatial accuracy, and controllability, outperforming prior baselines. This work represents the first dedicated diffusion framework for video bokeh generation, establishing a new baseline for temporally coherent and controllable depth-of-field effects. Code will be made publicly available.

Despite substantial progress in video understanding, most existing datasets are limited to Earth’s gravitational conditions. However, microgravity alters human motion, interactions, and visual semantics, revealing a critical gap for real-world vision systems. This presents a challenge for domain-robust video understanding in safety-critical space applications. To address this, we introduce MicroG-4M, the first benchmark for spatio-temporal and semantic understanding of human activities in microgravity. Constructed from real-world space missions and cinematic simulations, the dataset includes 4,759 clips covering 50 actions, 1,238 context-rich captions, and over 7,000 question–answer pairs on astronaut activities and scene understanding. MicroG-4M aims to support three core tasks: fine-grained multi-label action recognition, temporal video captioning, and visual question answering, thereby enabling a comprehensive evaluation of both spatial localization and semantic reasoning in microgravity contexts. We establish baselines using state-of-the-art models. All data, annotations, and code will be made publicly available upon decision.

Sparse attention mechanisms aim to reduce computational overhead with minimal accuracy loss by selectively processing salient tokens. Despite their effectiveness, most methods merely exploit a model’s inherent sparsity and thus plateau at moderate budgets (about 50\% token reduction), with little headroom to push budget lower without hurting accuracy. Other approaches attempt to enforce sparsity through trainable sparse attention or sharpness-inducing regularizers, but these either fix rigid patterns that ignore input and layer dynamics, or optimize proxy objectives without direct control over token budgets. In this paper, we explicitly reinforce token sparsity in well-posed multimodal large language models (MLLMs) through a simple RL-based post-training framework named $\textit{Sparsity Forcing}$. Our method explores the efficiency-accuracy trade-off by running multiple rollouts with different token budgets, where both efficiency (token reduction ratio) and performance (answer correctness) are formulated as joint rewards. By contrasting rollouts within each group, the more efficient and correct answer is rewarded while less efficient or incorrect ones are penalized, thereby turning token saving into an end-to-end, inference-consistent optimization objective. Across thirteen image and video benchmarks, Sparsity Forcing raises token reduction ratio on Qwen2-VL/Qwen2.5-VL from 20\% to 75\% with minimal accuracy decline, significantly reducing long-context inference memory by up to 3$\times$ while speeding up decoding by up to 3.3$\times$.

Dynamic Time Warping (DTW) is a classical method for measuring similarity between time series, but its non-differentiability hinders integration into end-to-end learning frameworks. To address this, soft-DTW replaces the minimum operator with a smooth soft-min, enabling differentiability and efficient computation. Motivated by soft-DTW, we propose perturbed-DTW, a differentiable framework of DTW obtained by adding random perturbations to warping costs and taking the expected minimum. Under Gumbel noise, perturbed-DTW exactly recovers soft-DTW, providing a natural probabilistic interpretation of soft-DTW. We further generalize this framework by extending the Gumbel noise to the broader family of generalized extreme value (GEV) distributions, leading to a new class of soft-DTW variants. Building on this insight, we introduce nested-soft-DTW (ns-DTW), which integrates GEV perturbations into the dynamic programming formulation of perturbed-DTW. This extension induces alignments with tunable skewness, offering greater flexibility in modeling diverse alignment structures. We validate ns-DTW on barycenter computation, clustering, and classification, demonstrating its effectiveness over existing approaches.

With the rapid adoption of diffusion models for visual content generation, proving authorship and protecting copyright have become critical. This challenge is particularly important when model owners keep their models private and may be unwilling or unable to handle authorship issues, making third-party verification essential. A natural solution is to embed watermarks for later verification. However, existing methods require access to model weights and rely on computationally heavy procedures, rendering them impractical and non-scalable. To address these challenges, we propose $\text{\emph{NoisePrints}}$, a lightweight watermarking scheme that utilizes the random seed used to initialize the diffusion process as a proof of authorship without modifying the generation process. Our key observation is that the initial noise derived from a seed is highly correlated with the generated visual content. By incorporating a hash function into the noise sampling process, we further ensure that recovering a valid seed from the content is infeasible. We also show that sampling an alternative seed that passes verification is infeasible, and demonstrate the robustness of our method under various manipulations. Finally, we show how to use cryptographic zero-knowledge proofs to prove ownership without revealing the seed. By keeping the seed secret, we increase the difficulty of watermark removal. In our experiments, we validate NoisePrints on multiple state-of-the-art diffusion models for images and videos, demonstrating efficient verification using only the seed and output, without requiring access to model weights.

Large language models (LLMs) have shown promise as parametric knowledge bases, but often underperform on question answering (QA) tasks due to hallucinations and uncertainty. While prior work attributes these failures to knowledge gaps in the model’s parameters, we uncover a complementary phenomenon: LLMs frequently retain correct knowledge even when generating incorrect or \``unsure'' answers. By analyzing the token-level output distributions, we find that correct answers often appear among high-probability candidates, despite not being selected. Motivated by this, we propose Hits@k, a novel metric to evaluate latent knowledge retention independent of answer surface form. Our experiments reveal that LLMs possess significantly more factual knowledge than is reflected by standard QA accuracy. Building on these insights, we further examine the prevailing few-shot QA paradigm. We find that prompting strategies which allow ``unsure'' outputs can inadvertently suppress correct answers by discouraging low-confidence generation. We design a set of quantitative experiments to measure this suppression effect, offering practical guidance for future prompt and decoding design in knowledge-intensive tasks.

Medical image segmentation is fundamental to clinical decision-making, yet existing models remain fragmented. They are usually trained on single knowledge sources and specific to individual tasks, modalities, or organs. This fragmentation contrasts sharply with clinical practice, where experts seamlessly integrate diverse knowledge: anatomical priors from training, exemplar-based reasoning from reference cases, and iterative refinement through real-time interaction. We present $\textbf{K-Prism}$, a unified segmentation framework that mirrors this clinical flexibility by systematically integrating three knowledge paradigms: (i) $\textit{semantic priors}$ learned from annotated datasets, (ii) $\textit{in-context knowledge}$ from few-shot reference examples, and (iii) $\textit{interactive feedback}$ from user inputs like clicks or scribbles. Our key insight is that these heterogeneous knowledge sources can be encoded into a dual-prompt representation: 1-D sparse prompts defining $\textit{what}$ to segment and 2-D dense prompts indicating $\textit{where}$ to attend, which are then dynamically routed through a Mixture-of-Experts (MoE) decoder. This design enables flexible switching between paradigms and joint training across diverse tasks without architectural modifications. Comprehensive experiments on 18 public datasets spanning diverse modalities (CT, MRI, X-ray, pathology, ultrasound, etc.) demonstrate that K-Prism achieves state-of-the-art performance across semantic, in-context, and interactive segmentation settings. Code is available at https://github.com/bangwayne/K-Prism.

Flow Matching (FM) has recently emerged as a powerful approach for high-quality visual generation. However, their prohibitively slow inference due to a large number of denoising steps limits their potential use in real-time or interactive applications. Existing acceleration methods, like distillation, truncation, or consistency training, either degrade quality, incur costly retraining, or lack generalization. We propose FlowCast, a training-free speculative generation framework that accelerates inference by exploiting the fact that FM models are trained to preserve constant velocity. FlowCast speculates future velocity by extrapolating current velocity without incurring additional cost, and accepts it if it is within a mean-squared error threshold. This constant-velocity forecasting allows redundant steps in stable regions to be aggressively skipped while retaining precision in complex ones. FlowCast is a plug-and-play framework that integrates seamlessly with any FM model and requires no auxiliary networks. We also present a theoretical analysis and bound the worst-case deviation between speculative and full FM trajectories. Empirical evaluations demonstrate that FlowCast achieves $>2.5\times$ speedup in image generation, video generation, and editing tasks, outperforming existing baselines with no quality loss as compared to standard full generation.