NeurIPS 2026 论文列表

Self-speculative decoding offers a plug-and-play route to accelerating autoregressive large language model (LLM) inference, but its practical success hinges on identifying effective layer-skipping configurations online without retraining or offline profiling. Existing methods mainly optimize agreement-based proxies that capture draft-target consistency, yet overlook intrinsic runtime differences across draft configurations. Consequently, configurations with similar acceptance can still yield substantially different realized decode throughput. We present SPIRIT, a speed-driven online adaptation framework for self-speculative decoding that directly optimizes realized decode throughput, measured as generated tokens per second, using window-level runtime feedback. Our key insight is that speed is not determined by agreement alone: it is jointly shaped by draft-target consistency and the intrinsic speed advantage of the draft configuration. Empirically, the skip ratio serves as a first-order control variable, capturing the dominant consistency-speed trade-off while sharply shrinking the search space over layer-skipping patterns. SPIRIT therefore first identifies an effective skip ratio and then refines the layer-skipping pattern within it. Combined with length-aware normalization and shift-aware profile-based adaptation, this design enables rapid adaptation to task shifts in non-stationary request streams. Across diverse tasks and model scales, SPIRIT consistently outperforms strong training-free baselines and achieves up to 1.74x speedup while preserving the target model's output distribution, establishing a practical foundation for plug-and-play, high-throughput LLM inference.

Recent foundation models are moving toward native multimodal Vision-Language Models (VLMs), making VLMs a central form of next-generation foundation models. However, their large language backbones make edge deployment difficult due to high memory footprint and memory-bound autoregressive decoding. Weight-only post-training quantization is a practical solution, but pushing VLMs to extreme low bit-widths remains challenging: existing rotation-free methods suffer from outliers at 2--3 bits, while rotation-based methods improve accuracy at the cost of additional runtime overhead. We propose SPHQuant, a rotation-free spherical weight-only quantization framework for VLMs. Instead of quantizing weights directly in Cartesian coordinates, SPHQuant decomposes each 8D weight vector into coordinate signs, radius, and a positive unit direction. This representation isolates outlier magnitude into the radius while keeping directions bounded and statistically regular. Based on this insight, SPHQuant allocates extra precision to the radius to mitigate accuracy degradation induced by outliers. It further uses a compact positive-direction codebook and fine-tunes codebook entries through angular parameterization to preserve the unit-sphere constraint. We also design a hardware-friendly GEMV kernel that keeps the direction codebook small enough for shared-memory lookup and packs radial bits efficiently. Experiments show that SPHQuant matches the performance of state-of-the-art extreme low-bit quantization methods while improving decode throughput by \textbf{30.3\%}, providing a new step toward efficient edge deployment of VLMs.

A fundamental open question in self-supervised learning (SSL) is the explicit characterization of the optimal geometry of the learned representations. Recently, LeJEPA identified isotropic Gaussian embeddings as optimal for minimizing downstream prediction risk in Euclidean spaces. However, the corresponding problem for distributions supported on lower-dimensional manifolds, such as the hypersphere, remains unexplored. In this work, we demonstrate that extending this minimax analysis to smooth distributions on Riemannian manifolds fundamentally changes the optimal solution. We show that, under a worst-case formulation, both \(k\)-nearest neighbors and kernel ridge regression induce hyperspherical uniformity. More precisely, we show that uniform distributions on manifolds are optimal for \(k\)-nearest neighbors, and that the uniform distribution on the sphere is optimal for kernel ridge regression with both the exponential dot-product kernel and the linear kernel. This theoretical insight reveals a fundamental limitation of Gaussian embeddings: their non-uniform density induces anisotropic $k$-NN neighborhoods, severely biasing the estimator. To correct this, we introduce \textbf{SPHERE-JEPA}, a theoretically grounded SSL framework. We adapt LeJEPA's Cramér--Wold projection mechanism to enforce hyperspherical uniformity rather than a Gaussian prior. Empirically, SPHERE-JEPA yields significant improvements, boosting texture retrieval mAP by over 6\%, while consistently matching or outperforming LeJEPA on standard benchmarks—including a $+1.8\%$ linear probing gain on ImageNet-1K (ViT-B/14).

Vision-Language Models for Earth Observation are bottlenecked at the visual encoder. Existing pipelines either compress multi-spectral signals into RGB pixels or collapse them into a single CLIP-style global vector, losing spectral information or spatial structure respectively. We show this is a consequence of coupling contrastive supervision to the dense feature stream, which degrades spectral fidelity. SPEXT resolves this through architectural decoupling. A wavelength-parameterized encoder produces multi-spectral patch tokens for dense tasks and VLM grounding, while a dedicated alignment token absorbs the contrastive gradient for retrieval. SPEXT simultaneously leads the PANGAEA segmentation benchmark (+2.51 mIoU), tops zero-shot retrieval among multi-spectral CLIP baselines (+4.6 mAP@100), and enables a frozen VLM, paired with a small projection adapter and LoRA, to answer free-form Earth-Observation questions directly from its tokens, substantially outperforming an RGB-pixel baseline on a downstream EO-RAG benchmark.

Scaling test-time compute has driven the recent advances in the reasoning capabilities of large language models (LLMs). However, increased compute often comes at the expense of higher user-facing latency, directly impacting user experience. Current test-time scaling methods primarily optimize for accuracy based on total compute resources (FLOPs), often overlooking latency constraints. To address this gap, we propose SPECS, a latency-aware test-time scaling method. SPECS uses a smaller, faster model to generate multiple candidates for each step of reasoning in parallel, and evaluates these candidates using a larger model and a quality critic. We design a theoretically grounded soft verification strategy to select a high-quality candidate to continue the generation. SPECS also employs a dynamic switching mechanism to use speculative drafts only for easier steps to maintain reasoning accuracy. Empirical results on a diverse and challenging set of reasoning and alignment benchmarks show that SPECS matches or surpasses the accuracy of SOTA test-time scaling methods while reducing latency by up to ~26%. Our theoretical analysis shows that as the amount of parallel compute scales, SPECS converges to an optimum of a KL-regularized reinforcement learning problem, a common objective for aligning LLM generation given a reward signal.

Stochastic Partial Differential Equations (SPDEs) driven by random noise play a central role in modeling physical processes with rough spatio-temporal dynamics, such as turbulence flows, superconductors, and quantum dynamics. Although machine learning (ML)-based surrogate models have shown promise for efficiently approximating such dynamics, progress remains limited by the lack of a unified benchmark with controlled data generation and comprehensive evaluation. This gap is particularly significant for singular SPDEs, for which benchmark datasets are largely unavailable and reliable simulation requires numerically delicate schemes based on renormalization. Moreover, subtle differences in data-generation procedures, such as noise approximation, basis choice, and the inclusion of renormalization, can significantly affect the resulting datasets and, consequently, model evaluation. We introduce SPDEBench, the first unified benchmark for ML-based SPDE learning. SPDEBench provides ready-to-use datasets for physically and mathematically significant SPDEs on 1-3D domains with periodic or Dirichlet boundary condition. Both regular and singular SPDEs are taken into consideration. SPDEBench also incorporates representative ML baselines in operator learning, together with 7 evaluation metrics, including Sobolev and distributional metrics beyond the standard $L^2$-error. Supported by SPDEBench, we conduct systematic evaluations of model accuracy, robustness, and out-of-distribution generalization under controlled data variations. Our numerical results show that SPDE-aware architectures generally achieve stronger performance than generic operator-learning baselines. These findings establish SPDEBench as a reproducible and extensible resource, paving pathway for principled benchmarking and architecture design for stochastic spatio-temporal dynamics.

Accurate epidemic forecasting is crucial for public health response, resource allocation, and outbreak intervention, but remains difficult with sparse, noisy, and highly non-stationary data. Because epidemics unfold across interacting regions, spatiotemporal methods are natural candidates for improving forecasts. Despite growing interest in spatial information, no standardized benchmark exists, and current evaluations often use simple chronological train-test splits that do not reflect real-time forecasting practice. We address this gap with SpatialEpiBench, a challenging benchmark for spatiotemporal epidemic forecasting in realistic public-health settings. SpatialEpiBench includes 11 epidemic datasets with standardized rolling evaluations and outbreak-specific metrics. We evaluate adjacency-informed forecasting models with widely used epidemic priors that adapt general models to epidemiology, but find that most methods underperform a simple last-value baseline from 1 day to 1 month ahead, even during outbreaks and with these priors. We identify three major failure modes: (1) poor outbreak anticipation, (2) difficulty handling sparsity and noise, and (3) limited utility of common geographic adjacency for epidemiological spatial information. We release benchmark data, code, and instructions at https://github.com/Rachel-Lyu/SpatialEpiBench to support development of operationally useful epidemic forecasting models.

Generalist neural routing solvers have shown great potential in solving diverse vehicle routing problems (VRPs) with a unified model. However, existing solvers are typically limited to symmetric settings or degrade in performance when switching to asymmetric settings due to input inconsistencies or inherent structural differences, substantially limiting their practicality in real-world scenarios that encompass both scenarios. To address this limitation, we define the spatial position of each node based on the relative distances to a specific set of pivots and further propose a Spatial Pivot-Aligned Coordinate-free Embedding (SPACE) framework that unifies node representation and solution generation across symmetric and asymmetric VRPs. Specifically, we construct a bidirectional Fréchet representation using a novel furthest pivot sampling strategy to enable invariant node representations across distinct problem settings. Furthermore, we introduce a weight-decomposed adaptive decoding mechanism that decouples geometric perception from problem representations, mitigating the overfitting of constraint decisions to a specific geometry setting. Extensive experiments on 110 VRP variants, comprising 55 symmetric problems and their asymmetric counterparts, demonstrate that SPACE achieves promising zero-shot generalization in both symmetric and asymmetric VRPs.

Monocular depth estimation (MDE) has advanced rapidly with the emergence of foundation models such as Depth Anything. Their transformer-based architectures provide strong generalization across diverse scenes and domains, but also incur high computational and memory cost, making efficient deployment challenging. Post-Training Quantization (PTQ) provides an efficient and practical solution for model compression, yet low-bit PTQ remains challenging for MDE models. When applying PTQ to Depth Anything, we identify two key challenges: (i) independently minimizing the quantization error of the query and key projections fails to preserve the attention maps induced by their interaction, and (ii) quantization errors accumulate across layers, resulting in intermediate feature distribution shifts. To address these issues, we propose SPA-Q, a Structure-Preserving Adaptive PTQ framework for MDE, consisting of two main components. First, Attention-Preserving Calibration (APC) determines query and key quantization parameters by matching the full-precision attention distribution. Second, Channel-Wise Distribution Alignment (CWDA) learns channel-wise affine transformations to mitigate quantization-induced distribution shifts, and the learned parameters are absorbed into the weights after training. Experimental results show that SPA-Q consistently outperforms existing PTQ methods under 4-bit quantization, achieving an average 25.9\% reduction in AbsRel and a 17.3\% improvement in $\delta_1$ across NYUv2 and KITTI datasets.

Collaborative learning is sustainable only if it benefits each participant; yet, standard Federated Learning (FL) optimizes a global average that often fails to serve individual clients. In heterogeneous settings, a client may rationally prefer training alone rather than contributing to a global model that targets "average-case" optimality but under performs locally. In this work, we address the problem of \emph{Selfish Personalization (SP)}: How can a target client leverage peer data to minimize its own risk? While current techniques often rely on heuristic performance proxies or clustering that lack sharp theoretical support, we propose \emph{SP-Convergence-Aware Client Weighting (SP-CACW)}. This novel framework determines the contribution of peer clients by explicitly minimizing a theoretical convergence bound for the target. By doing so, our approach efficiently separates useful signal from imported bias on-the-fly during the training process. We provide convergence guarantees that establish the theoretical superiority of \emph{SP-CACW}, alongside empirical results on the MNIST CIFAR datasets and LEAF Shakespeare.

In this paper, we introduce SP$^3$, a novel Plug-and-Play algorithm that accelerates Maximum \emph{a posteriori} image restoration by replacing denoisers with Spherical Encoders (SE) as generative priors. SP$^3$ approximates the intractable proximal prior step by utilizing the SE tightly structured latent space as a robust projection onto the natural image manifold. Alternating this projection with a closed-form data-consistency step, via Half-Quadratic Splitting, guarantees stable convergence without requiring gradient computation during inference. This unique formulation unlocks ``anytime'' restoration capabilities, producing sharp, plausible images from the first iteration. Evaluations across a variety of image restoration tasks demonstrate that SP$^3$ achieves perceptual quality comparable to state-of-the-art zero-shot diffusion and flow methods while being $3$-$630\times$ faster.

Speculative Decoding (SD) has shown strong success in accelerating large language model inference, but its efficiency remains limited for Vision-Language Models (VLMs). The diverse input modalities in VLMs make it difficult for a small trained drafter to achieve consistently high throughput across modalities. We address this limitation with an adaptive self-speculative decoding method. In particular, we adaptively select the skipped layers of VLMs to form the draft model that best fits the given query. Our selection algorithm SP$^2$ec utilizes the single-peak structure of the throughput with respect to the number of selected layers. Our algorithmic design explicitly minimizes the novel notion of wall-time regret for the given query, corresponding to the empirical wall-time latency. Theoretical guarantees on the wall-time regret show that SP$^2$ec is adaptive to the per-prompt single-peak structure. We conduct extensive experiments with Qwen3-VL and LLaVA-1.5 models across visual and textual tasks, which demonstrates the efficacy of the proposed SP$^2$ec.

Vision-language models (VLMs) have shown impressive capabilities across diverse tasks, motivating efforts to leverage these models to supervise robot learning. However, when used as evaluators in reinforcement learning (RL), today’s strongest models often fail under partial observability and distribution shift, enabling policies to exploit perceptual errors rather than solve the task. We introduce SOLE-R1 (Self-Observing LEarner), a video-language reasoning model explicitly designed to serve as the sole reward signal for online RL. Given only raw video observations and a natural-language goal, SOLE-R1 performs per-timestep spatiotemporal chain-of-thought (CoT) reasoning and produces dense estimates of task progress that can be used directly as rewards. To train SOLE-R1, we develop a large-scale video trajectory and reasoning synthesis pipeline that generates temporally grounded CoT traces aligned with continuous progress supervision. This data is combined with foundational spatial and multi-frame temporal reasoning, and used to train the model with a hybrid framework that couples supervised fine-tuning with RL from verifiable rewards. Across four different simulation environments and a real-robot setting, SOLE-R1 enables zero-shot online RL from random initialization: robots learn previously unseen manipulation tasks without ground-truth rewards, success indicators, demonstrations, or task-specific tuning. SOLE-R1 succeeds on 24 unseen tasks and substantially outperforms strong vision-language rewarders, including GPT-5 and Gemini-3-Pro, while exhibiting markedly greater robustness to reward hacking.

How fast *could* a deep-learning model run on target hardware, and how far is today’s implementation from that limit? These questions are central to software, hardware, and algorithm optimizations. Speed-of-Light (SOL) analysis answers them by computing a workload’s theoretical minimum execution time on a given architecture. Yet deriving SOL bounds remains manual, error-prone, and disconnected from rapid model development. To close this gap, we introduce SOLAR, a framework that automatically derives validated SOL bounds from Pytorch and JAX source code. SOLAR leverages both *generative* and *deterministic* components in its flow: an LLM frontend translates any source programs into an executable Affine Loop IR, validated by output comparison; a deterministic flow lifts the IR into an einsum graph; and an analytical backend computes unfused, fused, and cache-aware SOL bounds. SOLAR provides comprehensive operator and language coverage, produces validated bounds with zero observed SOL violations, and offers multi-fidelity analysis that tightens bounds and surfaces optimization insights. We evaluate SOLAR across KernelBench, JAX/Flax models, and robotics workloads. These experiments demonstrate four use cases: headroom analysis at multiple fidelity levels, identifying optimization opportunities, cross-platform exploration, and inverse-roofline hardware provisioning.

Vision Transformers dominate modern image benchmarks, however, what trained self-attention computes head by head remains unclear: prior work either injects inductive biases at initialization or clusters attention patterns visually without committing to a closed-form computation. We ask whether the softmax$(QK^\top)$ inside a trained head can be replaced, without further training, by a structured closed-form operator while preserving prediction and the residual-stream trajectory. We answer yes across ten ViTs and seven pretrainings (ImageNet-1k, ImageNet-21k FT, MAE, DINO, DINOv2, CLIP, SigLIP), and package the result as \textbf{SOLA}: a Structured Operator Library for Attention with three entries, each gated by a per-head diagnostic that upper bounds substitution error. The entries are a fixed 2D convolution kernel for shift heads, a per-image broadcast row for fixed-target heads, and a per-image rank-$k$ SVD of the attention matrix for near-global heads. A rank-adaptive extension picks $k$ per head from the SVD spectrum and extends structural coverage to every head on every backbone, with task-fidelity controlled by a single spectral threshold $\tau$ that trades compression for downstream fidelity. The diagnostics carry over to cross-attention: every decoder cross-attention head in BLIP captioning has effective rank $\approx 1$ and admits the broadcast substitution. A minimal reproducer that runs the full SOLA pipeline on a single backbone is included in the Supp. Mat. as a demo; the full codebase will be released open-source upon acceptance.

As agentic AI systems become increasingly capable of generating and optimizing GPU kernels, progress is constrained by benchmarks that reward speedup over software baselines rather than proximity to hardware-efficient execution. We present SOL-ExecBench, a benchmark of 235 CUDA kernel optimization problems extracted from 124 production and emerging AI models spanning language, diffusion, vision, audio, video, and hybrid architectures, targeting NVIDIA Blackwell GPUs. The benchmark covers forward and backward workloads across BF16, FP8, and NVFP4, including kernels whose best performance is expected to rely on Blackwell-specific capabilities. Unlike prior benchmarks that evaluate kernels primarily relative to software implementations, SOL-ExecBench measures performance against analytically derived Speed-of-Light (SOL) bounds computed by SOLAR, our pipeline for deriving hardware-grounded SOL bounds, yielding a fixed target for hardware-efficient optimization. We report a SOL Score that quantifies how much of the gap between a release-defined scoring baseline and the hardware SOL bound a candidate kernel closes. To support robust evaluation of agentic optimizers, we additionally provide a sandboxed harness with input correctness checks, isolated compilation and execution, CUPTI-based GPU activity timing, and defenses against common reward-hacking strategies. A third-party post-release evaluation by Cursor reported a $1.38\times$ geometric-mean speedup over a single-agent optimized PyTorch baseline and an overall SOL Score of 0.56 across the full benchmark. SOL-ExecBench reframes GPU kernel benchmarking from beating a mutable software baseline to closing the remaining gap to hardware Speed-of-Light.

Regression-based LiDAR relocalization has recently emerged as a promising solution for high-precision positioning in GNSS-denied environments. However, these methods are primarily tailored to autonomous driving, exhibiting significantly degraded accuracy in unmanned aerial vehicle (UAV) scenarios due to arbitrary pose variations and irregular flight paths. In this paper, we propose SOAR, a regression-based LiDAR relocalization framework for UAVs. Specifically, we introduce a locality-preserving sliding window attention module with locally invariant positional encoding to capture discriminative geometric structures robust to viewpoint changes. A coordinate-independent feature initialization module is further designed to eliminate sensitivity to global transformations. Furthermore, most existing UAV datasets are limited to evaluate LiDAR relocalization in real-world, due to the lack of synchronized LiDAR scans, accurate 6-DoF poses, or multiple traversals. Thus, we construct a large-scale UAV LiDAR localization dataset with 4 scenes and 13 irregular paths exhibiting rotation and altitude variations, providing a more realistic benchmark for UAVs. Extensive experiments demonstrate that our method achieves state-of-the-art performance, improving the localization success rate by 40\% and reducing mean error over 10m on UAVLoc. Our code and dataset will be released soon, and the dataset demo is available via the anonymous link.

We introduce SNACK, a string representation for probabilistic graph generation that enables sequence models to generate and model graph-structured data. SNACK linearizes graphs by encoding the nonzero entries of the lower-triangular adjacency matrix, establishing a bijective correspondence between valid sequences and ordered adjacency matrices. This design supports invalid logit masking for enforcing structural constraints, including chemical valence and ordering constraints, and enables principled string-based GFlowNet training by explicitly accounting for node-ordering symmetries. Across standard graph-generation tasks and molecular distribution-learning benchmarks, SNACK achieves competitive or state-of-the-art sample quality, while substantially improving the throughput of GFlowNet training relative to graph-based generation. Probing analyses further show that models trained on SNACK encode meaningful local and global graph structure despite receiving only sequential supervision. These results position SNACK as a practical framework for probabilistic graph generation with deep sequence models.

Multi-objective optimization aims to solve problems with competing objectives. Evaluating such problems is often slow or expensive, limiting the budget of evaluations. In many applications, historical data from related optimization tasks is available and can be leveraged via meta-learning to accelerate optimization. Bayesian optimization, as a promising technique for expensive black-box problems, has been extended independently to meta-learning and multi-objective optimization, but methods that simultaneously address both settings remain largely unexplored. We propose SMOG—a scalable and modular meta-learning model based on a multi-output Gaussian process—that explicitly learns correlations between objectives. SMOG builds a structured joint Gaussian process prior across meta- and target tasks and, after conditioning on metadata, yields a closed-form prior for the target task. This construction propagates metadata uncertainty into the target surrogate in a principled way. SMOG supports hierarchical, parallel training, achieving linear scaling with the number of meta-tasks. The resulting surrogate integrates seamlessly with standard multi-objective Bayesian optimization acquisition functions. We demonstrate that our method is consistently competitive, delivering strong data efficiency across representative benchmarks and applications.