Training-Free

LLM-as-a-Judge is a scalable alternative to human evaluation, yet existing rubric-based methods rely on human-annotated data such as reference answers or expert-crafted rubrics. We propose to automatically generate fine-grained evaluation rubrics without any human annotation. Our training-free method generates rubrics at dataset-specific and instance-specific granularities, achieving performance competitive with existing methods across four benchmarks. We further present a method that iteratively fine-tunes a rubric generator model via meta-judge reward signals. The fine-tuned generator outperforms all existing baselines in both pairwise and pointwise evaluation. Notably, a fine-tuned 14B rubric generator outperforms a much larger proprietary model at rubric generation, showing the effectiveness of our fine-tuning strategy.

Vision-language models (VLMs) rely on long visual token sequences for visual understanding, making the prefill stage expensive in both computation and memory. Most existing pruning methods follow an absolute-ranking paradigm, assigning importance scores to visual tokens and retaining a fixed top-K subset. In this work, we argue that this paradigm is fundamentally brittle: attention sinks distort token importance rankings, while image redundancy and query-dependent visual evidence make fixed token budgets unreliable across inputs. We propose OccamToken, a training-free framework that replaces absolute token ranking with register-anchored relative evidence testing. Instead of asking which tokens are globally important, OccamToken evaluates whether a visual token provides information beyond a register-based reference. Our key insight is that register tokens naturally absorb low-information attention patterns, making them a stable reference for identifying genuinely informative visual evidence. Based on this principle, OccamToken performs both image-adaptive redundancy pruning and query-adaptive relevance pruning through dynamic thresholds derived from register attention. Across LLaVA-NeXT, LLaVA-v1.5, and Qwen3-VL, OccamToken consistently improves the accuracy-efficiency trade-off without additional training. Notably, on LLaVA-NeXT, it reduces 2,880 visual tokens to approximately 40 while preserving over 93% of the original accuracy, enabling stable visual token compression even in the extreme 1.4% retention regime.

Reported retrieval scores for training-free shape descriptors conflate local signal design, normalization, aggregation, codebook fitting, and metric choices, making isolated component evaluation difficult. This paper reframes descriptor evaluation as a {\em protocol audit}. We introduce Diffused Geodesic Moments (DGM), a seed-conditioned descriptor that computes sparse implicit heat responses, converts them to distance-like fields, and summarizes each vertex by low-order moments across seeds and scales. DGM is used both as a practical non-spectral baseline and as an instrument for isolating protocol effects. On the registered FAUST benchmark split (FAUST-Reg) and the TOSCA shape collection, aggregation-matched experiments show that an independent Geometric Moment Shape Descriptor baseline built on Heat Kernel Signature features (GMSD-HKS) obtains the highest scores in this implementation ($0.621/0.820$ and $0.865/0.963$ mean average precision (mAP)/top-1), Wave Kernel Signature (WKS) remains a strong classical signal, and DGM is useful mainly when sparse solves, non-spectral deployment, or symmetry-informative seed frames are priorities. The broader finding is methodological: the input field and aggregation protocol can dominate the moment formula. The paper contributes a reproducible protocol-cascade analysis, a cross-shape alignment diagnostic for functional-map compatibility, and concrete recommendations for designing and reporting training-free shape descriptors.

Aligning LLMs for math tutoring typically requires RL-based training with multi-GPU infrastructure. We investigate whether training-free prompt optimization-evolving only the system prompt via API calls-can serve as a practical alternative. We adapt 7 published methods and propose 5 education-specialized methods, evaluating these 12 methods under 5 conditions on 2 OOD benchmark suites. All 12 best-per-method configurations surpass the strongest RL-trained baseline (R_total = 0.633), and our ParetoGrad achieves the best Pareto balance across post-test solve rate, leak control, and helpfulness, rather than dominating any single component. Behavioral analysis with an 82-code educational codebook reveals that training-free methods rely on teaching-knowledge patterns at 2-3x the rate of RL-trained models, with a compensating ~10 percentage-point reduction in intent-level scaffolding. We also find a task-dependent reasoning mode effect consistent across training-free and RL-based paradigms. Our approach enables efficient development of pedagogically aligned LLM tutors with prompts alone and minimal compute.

Style transfer with pre-trained diffusion models has advanced rapidly, but a core question remains underexplored: where in the model should style injection be strongest? StyleID, the leading training-free method, uses a single global parameter (gamma) uniformly across all layers and timesteps, which forces a fixed tradeoff between style quality and content preservation. We show this tradeoff is unnecessarily rigid. We systematically explore four dimensions of control: varying style injection strength across decoder layers, across denoising timesteps, and scheduling ControlNet geometric conditioning along both axes. The pattern is consistent everywhere: decreasing schedules, with stronger structural signal injection in shallower layers and earlier timesteps, reliably outperform the reverse. Beyond direction, schedule shape matters: cosine and square-root timestep schedules outperform linear. Most importantly, we find that gamma scheduling and ControlNet conditioning are nearly independent. The resulting combined configurations expand the Pareto frontier, offering superior tradeoffs between style fidelity and content preservation compared to any single baseline setting. Our best balanced configuration achieves ArtFID of 27.036 versus StyleID's 28.801 - a 6.1% relative improvement, with consistent gains across the full style-content tradeoff frontier. Results are validated across 35 configurations totaling over 28,000 stylized images using four complementary metrics. These findings generalize across SD backbones with identical rank ordering. All modifications are training-free, parameter-free, and require only a few lines of scheduling code; code is available at https://github.com/ameyskulkarni/scheduled_style_injection.

Audio-language models (ALMs) are increasingly used for audio captioning, question answering, and open-ended audio understanding, but their inference cost remains high when audio inputs are represented as long prefix-token sequences. These audio prefixes consume context budget, increase memory usage, and make deployment harder in resource-constrained or latency-sensitive settings. Existing training-free audio-token reduction methods mainly rely on fixed pooling or score-based pruning. Fixed pooling is content-agnostic, while score-based pruning can preserve isolated salient tokens but discard nearby acoustic context. We propose Local Temporal Bipartite Merging (LTBM), a training-free encoder-space compression method that merges similar nearby audio tokens under an explicit temporal window constraint. Beyond introducing LTBM, we use a controlled Global Merge variant to isolate whether temporal locality itself is a useful inductive bias for audio-token compression. Experiments on AudioCaps, Clotho, and MMAU with Qwen2-Audio show evidence of a task-dependent locality effect: locality-aware merging is more favorable for captioning at several compression settings, especially under stronger compression, while global matching is more competitive for multiple-choice audio understanding. A cross-backbone validation on Audio Flamingo 3 further supports the captioning-side advantage of locality-aware merging under moderate and aggressive compression.

Multi-shot video generation requires maintaining a consistent appearance of recurring entities across shots while remaining faithful to shot-specific text prompts. Recent autoregressive methods reuse previously generated frames as memory. However, full-frame storage entangles persistent entity information with transient scene context, leading to irrelevant information leakage and high computational cost. We propose an entity-centric memory in the form of an entity-indexed bank of latent patches. We introduce sparse token conditioning compatible with pretrained models, restricting self-attention to entity-relevant tokens and reducing computational cost. To support this, we introduce a structured multi-shot script format. We additionally propose a budgeted memory update strategy to maintain a compact, evolving memory. Finally, we equip the entity representation with a noise-injection mechanism that enables fine-grained appearance control, preventing leakage of irrelevant information. Our method improves prompt adherence and efficiency while preserving subject consistency.

Although existing video editing methods are generally feasible, they often require many costly iterations and still struggle to deliver high-quality yet satisfying editing results. We attribute this limitation to the prevalent data-to-data paradigm, which is less compatible with modern generative models than noise-to-data generation. To address this gap, we revisit video editing from a noise-to-data perspective and propose Streaming-Generation-based Video Editing (StreamEdit), which preserves few-step sampling while seamlessly injecting source-video conditions. Built on pre-trained streaming generation models, StreamEdit introduces dual-branch fast sampling with a self-attention bridge and cross-attention grounding/boosting to satisfy both sampling and conditioning requirements. We further propose source-oriented guidance to improve target-generation quality, and a visual prompting strategy to enhance editing flexibility and practicality. The method is effective, robust, and generalizable across different models. Extensive experiments on diverse video editing tasks show that StreamEdit consistently outperforms existing approaches, even in few-step settings with minimal time cost. Code and results are available at: https://dsl-lab.github.io/StreamEdit/.

Training-free zero-shot composed image retrieval models are recently gaining increasing research interest due to their generalizability and flexibility in unseen multimodal retrieval. Recent LLM-based advances focus on generating the expected target caption by exploring the compositional ability behind the LLMs. Although efficient, we find that 1) the generated captions tend to introduce unexpected features from the reference image due to the semantic gap between the input image and text modification, where the image contains much more details than the text; 2) the point-to-point alignment during the retrieval stage fails to capture diverse compositions. To address these challenges, we introduce a novel Semantic Transition and Transportation in collaboration framework for training-free zero-shot CIR tasks. Specifically, given the composed caption inferred by an LLM, we aim to refine it through a transition vector in the embedding space and make it closer to the target image. Combining LLMs with user instruction, the refined caption concentrates more on the core modification intent and thus filters out unnecessary noise. Moreover, to explore diverse alignment during the retrieval stage, we model the caption and image as discrete distributions and reformulate the retrieval task as a set-to-set alignment task. Finally, a bidirectional transportation distance is developed to consider fine-grained alignments across modalities and calculate the retrieval score. Extensive experiments demonstrate that our method can be general, effective, and beneficial for many CIR tasks.

We propose STEAM (Spatial, Temporal, and Emergent congestion Awareness for MAPF), a training-free test-time enhancement framework for learning-based decentralized Multi-Agent Path Finding (MAPF) in discrete environments. Given a pretrained decentralized policy, STEAM requires no retraining, architectural modification, or replacement by a centralized planner. Instead, it injects lightweight congestion-aware guidance into the original policy execution. STEAM first rolls out the shortest paths induced by the current cost-to-go maps to identify potential future congestion hotspots. Spatially avoidable congestion is mitigated by updating agent-specific cost-to-go information, while spatially unavoidable bottlenecks are handled through temporal logit correction. In addition, emergent local congestion is reduced by a density-aware logit correction based on neighboring agents' corrected cost-to-go maps. Extensive experiments on representative learning-based decentralized MAPF algorithms show that STEAM consistently improves success rate, makespan, and solution cost, with success-rate gains of up to 60% and only minor computational overhead. The implementation is available at https://anonymous.4open.science/r/STEAM-MAPF-7A62.

Graph Anomaly Detection (GAD) is increasingly shifting to Generalist GAD (GGAD) for cross-domain "one-for-all" detection, but existing GGAD methods predominantly rely on the neighbor consistency principle, falling into the \textbf{Node-to-Neighbor Consistency Paradigm} for anomaly quantification. These methods suffer from complex training pipelines, heavy training data dependency, high computational costs, and unstable cross-domain generalization. To address these limitations, we propose NeighborDiv, a training-free generalist graph anomaly detection framework based on neighbor diversity. Departing from the dominant Node-to-Neighbor Consistency Paradigm, we shift the focus to the \textbf{Neighbor-to-Neighbor Diversity Paradigm}, and uncover that the internal structural dispersion of a node's neighbor set is a powerful, independently discriminative anomaly signal. We quantify neighbor diversity via the variance of inter-neighbor feature similarities, which captures how a node organizes its local graph environment, and operates independently of conventional node-to-neighbor consistency frameworks. Extensive experiments under two standard GGAD evaluation paradigms show NeighborDiv achieves state-of-the-art performance, with relative gains of 10.25% in average AUC and 17.78% in average AP over the second-best baseline under Single-Domain Independent Training (SDIT), and 6.89%/9.58% in AUC/AP under Unified Multi-Domain Training (UMDT), respectively. Notably, NeighborDiv yields zero performance volatility across all datasets, eliminating training-set dependency and establishing a lightweight and highly practical GGAD framework.

Without incurring significant computational overhead, train-free long video generation aims to enable foundation video generation models to produce longer videos. Frame-level autoregressive frameworks, e.g., FIFO-diffusion, offer the advantage of generating infinitely long videos with constant memory consumption. However, the mismatch between training and inference, coupled with the challenge of maintaining long-term consistency, limits the effective utilization of foundation models. To mitigate these concerns, we propose \textbf{MIGA}, a novel infinite-frame long video generation method. Firstly, we propose an effective two-stage alignment mechanism that mitigates the training-inference gap by reducing the excessive noise span fed to the model. We then introduce an innovative dual consistency enhancement mechanism, where the self-reflection approach corrects early high-noise frames and the long-range frame guidance approach leverages later low-noise frames with broad coverage to steer generation, jointly improving temporal consistency. Extensive experiments on VBench and NarrLV demonstrate the state-of-the-art performance of MIGA. Our project page is available at https://xiaokunfeng.github.io/miga_homepage/.

Open-vocabulary segmentation models such as SAM3 perform well across broad categories via text prompting, yet degrade when target classes are visually underrepresented in pretraining or depart from canonical depictions-limitations text prompts cannot resolve spatially. We present SegRAG, a training-free retrieval-augmented segmentation framework that grounds SAM3 with class-specific point prompts derived from a curated DINOv3 feature bank. Offline, dense patch-level descriptors are extracted from annotated references and filtered by Intra-Class Cohesion Distillation (ICCD), retaining only prototypes that reliably retrieve within-class foreground. At inference, Topographic Similarity Grounding (TSG) computes a cosine-similarity landscape against retrieved prototypes, identifies coherent high-confidence regions via connected-component analysis, and extracts peak locations through non-maximum suppression. The resulting point prompts are delivered jointly with class-name text in a single SAM3 forward pass. On four standard benchmarks, SegRAG consistently outperforms the text-only baseline, gaining up to +3.92 mIoU on LVIS. On AgML agricultural benchmarks under zero-shot domain transfer, it raises mean IoU from 25.27 to 59.24 (+33.97) and recovers individual classes from zero to over 95 mIoU. Ablations confirm that ICCD, TSG, and joint prompting each contribute independently and compound when combined. Code is available at (https://github.com/boudiafA/SegRAG).

Recent studies introduce conditional memory modules that decouple knowledge storage from neural computation, enabling more direct knowledge access. Compared to MoE, which relies on dynamic computation paths, explicit lookup provides a more efficient knowledge retrieval mechanism. However, these approaches still depend on learned memory embeddings, requiring additional training and limiting flexibility. To address this, we propose N-gram Memory (NGM), a training-free, plug-and-play module composed of a Causal N-Gram Encoder and a Cosine-Gated Memory Injector. The Causal N-Gram Encoder directly averages the pretrained token embeddings of the backbone model to construct N-gram representations, thereby eliminating the need to train separate N-gram embeddings from scratch. This design requires neither an additional memory table nor a retrieval pipeline. The Cosine-Gated Memory Injector then uses a non-parametric cosine gate with ReLU to modulate the retrieved embeddings into the contextual representations. We evaluate NGM on the Qwen3 series from 0.6B to 14B across eight benchmarks. NGM improves average performance by 0.5 to 1.2 points, with particularly clear gains on code generation and knowledge-intensive tasks (e.g., +3.0 on LiveCodeBench and +3.03 on GPQA for Qwen3-14B). Moreover, NGM also improves performance in multimodal benchmarks (e.g., MMStar +1.53 on Qwen3-VL-2B).

Planning from raw visual input remains a significant challenge for current Vision-Language Models (VLMs), when the complexity of input is beyond their one-step perception capability. Motivated by recent advances in Thinking with Images (TWI), a reasonable solution is to decompose the perception process into simpler steps by iteratively acquiring and incorporating local visual evidence. However, even though current VLMs are well-trained in general TWI ability, their perceptual bottleneck in the planning domain remains. To tackle this challenge, we formulate TWI as a tool to gradually build and reflect an accurate internal world model. We find that the resulting training-free planning strategy enables VLMs to solve tasks that are far beyond their initial capabilities, at the cost that too many TWI operations would significantly increase the computational overhead. To further improve efficiency, we propose Pattern Inference, a novel TWI strategy enabling VLMs to actively recognize known visual patterns in the new tasks and directly infer local world model structures. To obtain these patterns, we propose Pattern Induction, an online inductive learning strategy treating visual patterns as composite and reusable experts, which are autonomously discovered and optimized from experience. Experimental evaluations in FrozenLake, Crafter and CubeBench domains show that our approaches achieve a desirable balance between accuracy and efficiency.

Visual anomaly detection (AD) for industrial inspection is a highly relevant task in modern production environments. The problem becomes particularly challenging when training and deployment data differ due to changes in acquisition conditions during production. In the VAND 4.0 Industrial Track, models must remain robust under distribution shifts such as varying illumination and their performance is assessed on the MVTec AD 2 dataset. To address this setting, we propose a training-free and class-agnostic anomaly detection pipeline based on the work of SuperAD. Our approach improves generalization through several modifications designed to enhance robustness under distribution shifts. These adaptations include using a DINOv3 backbone, overlapping patch-wise processing, intensity-based augmentations, improved memory-bank subsampling for better coverage of the data distribution, and iterative morphological closing for cleaner and more spatially consistent anomaly maps. Unlike methods that rely on class-specific architectures or per-class hyperparameter tuning, our method uses a single architecture and one shared hyperparameter configuration across all object classes. This makes the approach well suited for industrial deployment, where product variants and appearance changes must be handled with minimal adaptation effort. We achieve segmentation F1 scores of $62.61\%$, $57.42\%$, and $54.35\%$ on test public, private, and private mixed of MVTec AD 2 respectively, thereby outperforming SuperAD and other state-of-the-art methods. Code is available at https://github.com/LukasRoom/SuperADD.

Recent text-to-video diffusion transformers generate visually compelling frames, yet still struggle with temporal coherence, often producing flickering, drifting, or unstable motion. We show that these failures leave a clear imprint inside the model: incoherent videos consistently exhibit irregular, fragmented temporal diagonals in their intermediate self-attention maps, whereas stable motion corresponds to smooth, band-diagonal patterns. Building on this observation, we introduce TeDiO, a training-free, inference-time method that reinforces temporal consistency by regularizing these internal attention patterns. TeDiO estimates diagonal smoothness, identifies unstable regions, and performs lightweight latent updates that promote coherent frame-to-frame dynamics, without modifying model weights or using external motion supervision. Across multiple video diffusion models (e.g., Wan2.1, CogVideoX), TeDiO delivers markedly smoother motion while preserving per-frame visual quality, offering an efficient plug-and-play approach to improving dynamic realism in modern video generation systems.

In large vision-language models, visual tokens typically constitute the majority of input tokens, leading to substantial computational overhead. To address this, recent studies have explored pruning redundant or less informative visual tokens for image understanding tasks. However, these methods struggle with pixel grounding tasks, where token importance is highly contingent on the input text. Through an in-depth analysis of CLIP, we observe that visual tokens within referent regions often exhibit low similarity to their textual representation. Motivated by this insight, we introduce LiteLVLM, a training-free, text-guided token pruning strategy for efficient pixel grounding inference. By reversing the ranking of CLIP's visual-text similarity, LiteLVLM effectively retains visual tokens covering the referent regions, while recovering context tokens to enable clear foreground-background separation. Extensive experiments demonstrate that LiteLVLM significantly outperforms existing methods by over 5% across diverse token budgets. Without any training or fine-tuning, LiteLVLM maintains 90% of the original performance with a 22% speedup and a 2.3X memory reduction. Our code is available at https://github.com/sejong-rcv/LiteLVLM.

Language-guided segmentation transcends the scope limitations of traditional semantic segmentation, enabling models to segment arbitrary target regions based on natural language instructions. Existing approaches typically adopt a two-stage framework: employing Multimodal Large Language Models (MLLMs) to interpret instructions and generate visual prompts, followed by foundational segmentation models (e.g., SAM) to produce masks. However, due to the limited spatial grounding capabilities of off-the-shelf MLLMs, these methods often rely on extensive training on large-scale datasets to achieve satisfactory accuracy. While recent advances have introduced reasoning mechanisms to improve performance, they predominantly operate within the textual domain, performing chain-of-thought reasoning solely based on abstract text representations without direct visual feedback. In this paper, we propose Seg-Agent, a completely training-free framework that pioneers Explicit Multimodal Chain-of-Reasoning. Unlike prior text-only reasoning, our approach constructs an interactive visual reasoning loop comprising three stages: generation, selection, and refinement. Specifically, we leverage Set-of-Mark (SoM) visual prompting to render candidate regions directly onto the image, allowing the MLLM to ``see'' and iteratively reason about spatial relationships in the visual domain rather than just the textual one. This explicit multimodal interaction enables Seg-Agent to achieve performance comparable to state-of-the-art training-based methods without any parameter updates. Furthermore, to comprehensively evaluate generalization across diverse scenarios, we introduce Various-LangSeg, a novel benchmark covering explicit semantic, generic object, and reasoning-guided segmentation tasks. Extensive experiments demonstrate the effectiveness and robustness of our method.

Distilled one-step (T=1) or few-step (T$\leq$4) diffusion models enable real-time image generation but often exhibit reduced sample diversity compared to their multi-step counterparts. In multi-step diffusion, diversity can be introduced through schedules, trajectories, or iterative optimization; however, these mechanisms are unavailable in the few-step or single-step setting, limiting the effectiveness of existing diversity-enhancing methods. A natural alternative is to perturb intermediate features, but naive feature perturbation is often ineffective, either yielding limited diversity gains or degrading generation quality. We argue that effective diversity injection in few-step models requires perturbations that respect the model's learned feature geometry. Based on this insight, we propose STRIDE, a training-free and optimization-free method that operates in a single forward pass. STRIDE injects spatially coherent (pink) noise into intermediate transformer features, projected onto the principal components of the model's own activations, ensuring that perturbations lie on the learned feature manifold. This design enables controlled variation along meaningful directions in the representation space. Extensive experiments on FLUX.1-schnell and SD3.5 Turbo across COCO, DrawBench, PartiPrompts, and GenEval show that STRIDE consistently improves diversity while maintaining strong text alignment. In particular, STRIDE reduces intra-batch similarity with minimal impact on CLIP score, and Pareto-dominates existing training-free baselines on the diversity-fidelity frontier. These results highlight that, in the absence of iterative refinement, improving diversity in few-step and one-step diffusion depends not on increasing perturbation strength, but on aligning perturbations with the model's internal representation structure.

Vision-Language-Action (VLA) models achieve remarkable flexibility and generalization beyond classical control paradigms. However, most prevailing VLAs are trained under a single-frame observation paradigm, which leaves them structurally blind to temporal dynamics. Consequently, these models degrade severely in non-stationary scenarios, even when trained or finetuned on dynamic datasets. Existing approaches either require expensive retraining or suffer from latency bottlenecks and poor temporal consistency across action chunks. We propose Pace-and-Path Correction, a training-free, closed-form inference-time operator that wraps any chunked-action VLA. From a single quadratic cost, joint minimization yields a unified solution that decomposes orthogonally into two distinct channels. The pace channel compresses execution along the planned direction, while the path channel applies an orthogonal spatial offset, jointly absorbing the perceived dynamics within the chunk window. We evaluate our approach on a comprehensive diagnostic benchmark MoveBench designed to isolate motion as the sole controlled variable. Empirical results demonstrate that our framework consistently outperforms state-of-the-art training-free wrappers and dynamic-adaptive methods and improves success rates by up to 28.8% and 25.9% in absolute terms over foundational VLA models in dynamic-only and static-dynamic mixed environments, respectively.

Identity-preserved image generation is typically built on many-step diffusion backbones, making personalized generation expensive at deployment time. We show that this cost is often unnecessary for identity-conditioned FLUX generation. A frozen InfuseNet identity adapter trained with dev transfers directly to the distilled schnell backbone without retraining. This two-line replacement -- changing the backbone path and disabling classifier-free guidance -- reduces latency by 5.9x while improving ArcFace identity similarity by +0.028 and lpips by -0.016 over the standard 28-step dev baseline. To explain why this works, we analyze the denoising trajectory and find that identity fidelity enters an early effective regime, often within 4-8 steps, while later steps primarily refine visual detail, sharpness, and contrast. Adapter ablations confirm that identity formation depends on the identity adapter, while attention-stream norm probes suggest that the relative conditioning contribution decreases as sampling proceeds. Preliminary style-adapter and object-adapter sweeps on SDXL and SD1.5 show similar diminishing returns after intermediate steps. These results position distilled backbone replacement as a simple, training-free strategy for improving the efficiency-fidelity tradeoff of identity-preserved generation.

Existing training-free approaches for GUI grounding often rely on multiple inference runs, such as iterative cropping or candidate aggregation, to identify target elements. Despite this additional computation, each forward pass still independently interprets the instruction and parses the visual layout, without enabling progressive interaction among visual tokens. In this paper, we study what happens during GUI grounding in Vision-Language Models (VLMs) and identify a previously overlooked bottleneck. We show that grounding follows a two-stage paradigm: the prefill stage determines candidate UI elements, while the decoding stage subsequently refines the final coordinates. This asymmetry establishes prefill as the critical step, as errors in candidate selection cannot be effectively corrected during decoding. Based on this observation, we propose Re-Prefill, a training-free method that revisits inference by introducing an attention-guided second prefill stage to refine target selection. Specifically, visual tokens that consistently receive high attention from the query position, i.e., the final token, across layers are extracted as a preliminary target hypothesis and appended to the input, together with the instruction hidden states, enabling the model to deeply re-think its decision before coordinate generation. Experiments across four VLMs and five benchmarks, including ScreenSpot-Pro, ScreenSpot-V2, OSWorld-G, UI-Vision, and MMBench-GUI, demonstrate consistent improvements without additional training, with gains of up to 4.3% on ScreenSpot-Pro. Code will be available at https://github.com/linjiaping1/Re-Prefill.

Stable Diffusion (SD) has evolved DDPM (Denoising Diffusion Probabilistic Model) based image generation significantly by denoising in latent space instead of feature space. This popularized DDPM-based image generation as the cost and compute barrier was significantly lowered. However, these models could only generate fixed-resolution images according to their training configuration. When we attempt to generate higher resolutions, the resulting images show object duplication artifacts consistently. To solve this problem without finetuning SD models, recent works have tried dilating the convolution kernels of the models and have achieved a great level of success. But dilated kernels are harder to fine-tune due to being zero-gapped. Apart from this, other methods, such as patched diffusion, could not solve the object-duplication problem efficiently. Hence, to overcome the limitations of dilated convolutions, we propose kernel interpolation of SD models for higher-resolution image generation. In this work, we show mathematically that interpolation can correctly scale convolution kernels if multiplied by a constant coefficient and achieve competitive empirical results in generating beyond-training-resolution images with Stable Diffusion using zero training. Furthermore, we demonstrate that our method enables interpolation of deep neural networks to adapt to higher-dimensional training data, with a worst-case performance drop of $2.6\%$ in accuracy and F1-Score relative to the baseline. This shows the applicability of our method to be general, where we interpolate fully-connected layers, going beyond convolution layers. We also discuss how we can reduce the memory footprints of training neural networks, using our method up to at least $4\times$.

Training-free guidance enables pre-trained diffusion and flow models to optimize application-specific objectives using feedback from external black-box reward functions. However, existing methods are feedback-inefficient because reward feedback is used only transiently to inform a localized gradient approximation or a discrete search decision, and is subsequently discarded. To address this limitation, we propose Flow-Direct, a framework that guides the generation process via a persistent guidance field. Theoretically, this guidance field is analytically derived from the log-density ratio between the base and reward-weighted target distributions; it transports the pre-trained distribution to the target distribution. In practice, the field is implemented as a non-parametric estimator constructed from all accumulated reward-evaluated samples. As more samples are collected during optimization, this empirical guidance field becomes increasingly accurate. This persistent formulation yields two major advantages. First, Flow-Direct is highly feedback-efficient: because every evaluated sample is used to refine the global guidance field, no reward information is wasted. Second, the framework is naturally reusable: once optimization is complete, the collected dataset defines a reusable guidance field for generating novel target samples without additional reward evaluations, and distinct guidance fields can be combined to generate samples that simultaneously satisfy multiple objectives.

Dense hand contact estimation requires both high-level semantic understanding and fine-grained geometric reasoning of human interaction to accurately localize contact regions. Recently, multi-modal large language models (MLLMs) have demonstrated strong capabilities in understanding visual semantics, enabled by vision-language priors learned from large-scale data. However, leveraging MLLMs for dense hand contact estimation remains underexplored. There are two major challenges in applying MLLMs to dense hand contact estimation. First, encoding explicit 3D hand geometry is difficult, as MLLMs primarily operate on vision and language modalities. Second, capturing fine-grained vertex-level contact remains challenging, as MLLMs tend to focus on high-level semantics rather than detailed geometric reasoning. To address these challenges, we propose ContactPrompt, a training-free and zero-shot approach for dense hand contact estimation using MLLMs. To effectively encode 3D hand geometry, we introduce a detailed hand-part segmentation and a part-wise vertex-grid representation that provides structured, localized geometric information. To enable accurate and efficient dense contact prediction, we develop a multi-stage structured contact reasoning with part conditioning, progressively bridging global semantics and fine-grained geometry. Therefore, our method effectively leverages the reasoning capabilities of MLLMs while enabling precise dense hand contact estimation. Surprisingly, the proposed approach outperforms previous supervised methods trained on large-scale dense contact datasets without requiring any training. The codes will be released.

Diffusion-based image editing offers strong semantic controllability, but remains computationally expensive due to iterative high-resolution denoising over all spatial tokens. Dynamic-resolution sampling reduces this cost by performing early steps at reduced resolution. However, existing approaches prioritize upsampling using low-level heuristics such as edge detection or channel variance, which are weakly aligned with editing semantics and may lead to structural inconsistency. Moreover, spatial regions are often upsampled without verifying whether semantic modification is actually required, resulting in redundant high-resolution computation and accumulated errors. Therefore, we propose SpecEdit, a training-free dynamic-resolution framework tailored for diffusion-based image editing. SpecEdit follows a draft-and-verify scheme: a low-resolution draft first estimates the semantic outcome, after which token-level discrepancies are used to identify edit-relevant tokens for high-resolution denoising, while the remaining tokens stay at a coarse resolution. Experiments on Qwen-Image-Edit and FLUX.1-Kontext-dev demonstrate up to 10x and 7x acceleration, while maintaining strong quality. SpecEdit is complementary to step distillation and other acceleration techniques, achieving up to 13x speedup when combined with existing methods. Our code is in supplementary material and will be released on GitHub.

We present a training-free approach for controllable 3D inpainting based on initial noise optimization. In the structured 3D latent diffusion framework, we observe that the underlying geometric structure is established during the early stages of the diffusion process and exhibits high sensitivity to the initial noise. Such characteristics compromise stability in tasks like inpainting and editing, where the model must ensure strict alignment with the existing context while synthesizing a new structure. In this paper, we introduce a strategy to optimize the initial noise within the structured 3D latent diffusion framework, ensuring high-fidelity 3D inpainting. Specifically, we update the initial noise by leveraging a backpropagation approximation grounded in the rectified flow model, with the spectral parameterization specially designed for robust and efficient structured 3D latent optimization. Experiments demonstrate consistent improvements in contextual consistency and prompt alignment over representative training-free inpainting baselines, establishing initial noise control as an independent dimension for 3D inpainting, orthogonal to conventional sampling trajectory manipulation.

I propose the \emph{Random Cloud} method, a training-free approach to neural architecture search that discovers minimal feedforward network topologies through stochastic exploration and progressive structural reduction. Unlike post-training pruning methods that require a full train-prune-retrain cycle, this method evaluates randomly initialized networks without backpropagation, progressively reduces their topology, and only trains the best minimal candidate at the end. I evaluate on 7 classification benchmarks against magnitude pruning and random pruning baselines. The Random Cloud matches or outperforms both baselines in 6 of 7 datasets, achieving statistically significant improvements on Sonar ($+4.9$pp accuracy, $p{=}0.017$ vs magnitude pruning) with 87% parameter reduction. Crucially, the method is faster than both pruning baselines in 4 of 5 datasets (0.67--0.94$\times$ the cost of full training), since it avoids training the full-size network entirely.

Transformer-based architectures have established a dominant paradigm in global semantic perception; however, they remain fundamentally constrained by the profound spatial heterogeneity inherent in natural images. Specifically, the imposition of a uniform global receptive field across regions of varying information density inevitably leads to local feature degradation, particularly in dense conflict zones populated by microscopic targets. To address this mechanistic limitation, we propose ViCrop-Det, a training-free inference framework that introduces adaptive spatial trust region shrinkage. Inspired by the use of attention entropy in anomaly segmentation, ViCrop-Det leverages the detection decoder's cross-attention distribution as an endogenous probe. By utilizing Spatial Attention Entropy (SAE) to heuristically evaluate local spatial ambiguity, the framework executes dynamic spatial routing, allocating a fixed computational budget exclusively to regions exhibiting both high target saliency and high cognitive uncertainty. By shrinking the spatial trust region and injecting high-frequency localized observations, ViCrop-Det actively resolves spatial ambiguity and recovers fine-grained features without requiring architectural modifications. Extensive evaluations on VisDrone and DOTA-v1.5 demonstrate that ViCrop-Det yields competitive performance enhancements, consistently adding +1-3 mAP@50 to RT-DETR-R50 and Deformable DETR with a marginal 20-23% latency overhead. On MS COCO, $AP_{S}$ improves while $AP_{M}/AP_{L}$ remains stable, indicating precise fine-scale refinement without compromising the global spatial prior. Under compute-matched settings, our adaptive routing strategy comprehensively surpasses uniform slicing baselines, achieving a highly optimized accuracy-speed trade-off.

Open-vocabulary change detection aims to identify semantic changes in bi-temporal remote sensing images without predefined categories. Recent methods combine foundation models such as SAM, DINO and CLIP, but typically process each timestamp independently or interact only at the final comparison stage. Such paradigms suffer from insufficient temporal coupling during semantic reasoning, which limits their ability to distinguish genuine semantic changes from non-semantic appearance discrepancies. In addition, patch-dominant inference on high-resolution images often weakens global semantic continuity and produces fragmented change regions. To address these issues, we propose MemOVCD, a training-free open-vocabulary change detection framework based on cross-temporal memory reasoning and global-local adaptive rectification. Specifically, we reformulate bi-temporal change detection as a two-frame tracking problem and introduce weighted bidirectional propagation to aggregate semantic evidence from both temporal directions. To stabilize memory propagation across large temporal gaps, we construct histogram-aligned transition frames to smooth abrupt appearance changes. Moreover, a global-local adaptive rectification strategy adaptively fuses local and global-view predictions, improving spatial consistency while preserving fine-grained details. Experiments on five benchmarks demonstrate that MemOVCD achieves favorable performance on two change detection tasks, validating its effectiveness and generalization under diverse open-vocabulary settings.

Long-term video understanding requires interpreting complex temporal events and reasoning over procedural activities. While instructional video corpora, like HowTo100M, offer rich resources for model training, they present significant challenges, including noisy ASR transcripts and inconsistent temporal alignments between narration and visual content. In this work, we introduce an automated, training-free pipeline to extract high-quality procedural annotations from in-the-wild instructional videos. Our approach segments videos into coherent shots, filters poorly aligned content, and leverages state-of-the-art multimodal and large language models (Qwen2.5-VL and DeepSeek-R1) to generate structured, temporally grounded procedural steps. This pipeline yields DenseStep2M, a large-scale dataset comprising approximately 100K videos and 2M detailed instructional steps, designed to support comprehensive long-form video understanding. To rigorously evaluate our pipeline, we curate DenseCaption100, a benchmark of high-quality, human-written captions. Evaluations demonstrate strong alignment between our auto-generated steps and human annotations. Furthermore, we validate the utility of DenseStep2M across three core downstream tasks: dense video captioning, procedural step grounding, and cross-modal retrieval. Models fine-tuned on DenseStep2M achieve substantial gains in captioning quality and temporal localization, while exhibiting robust zero-shot generalization across egocentric, exocentric, and mixed-perspective domains. These results underscore the effectiveness of DenseStep2M in facilitating advanced multimodal alignment and long-term activity reasoning. Our dataset is available at https://huggingface.co/datasets/mingjige/DenseStep2M.

Long-context large language models remain computationally expensive to run and often fail to reliably process very long inputs, which makes context compression an important component of many systems. Existing compression approaches typically rely on trained compressors, dense retrieval-style selection, or heuristic trimming, and they often struggle to jointly preserve task relevance, topic coverage, and cross-sentence coherence under a strict token budget. To address this, we propose a training-free and model-agnostic compression framework that selects a compact set of sentences guided by structural graph priors. Our method constructs a sparse hybrid sentence graph that combines mutual k-NN semantic edges with short-range sequential edges, extracts a topic skeleton via clustering, and ranks sentences using an interpretable score that integrates task relevance, cluster representativeness, bridge centrality, and a cycle coverage cue. A budgeted greedy selection with redundancy suppression then produces a readable compressed context in original order. Experimental results on four datasets show that our approach is competitive with strong extractive and abstractive baselines, demonstrating larger gains on long-document benchmarks.

Vision-Language Models (VLMs) have become central to autonomous driving systems, yet their deployment is severely bottlenecked by the massive computational overhead of multi-view camera and multi-frame video input. Existing token pruning methods, primarily designed for single-image inputs, treat each frame or view in isolation and thus fail to exploit the inherent spatio-temporal redundancies in driving scenarios. To bridge this gap, we propose ST-Prune, a training-free, plug-and-play framework comprising two complementary modules: Motion-aware Temporal Pruning (MTP) and Ring-view Spatial Pruning (RSP). MTP addresses temporal redundancy by encoding motion volatility and temporal recency as soft constraints within the diversity selection objective, prioritizing dynamic trajectories and current-frame content over static historical background. RSP further resolves spatial redundancy by exploiting the ring-view camera geometry to penalize bilateral cross-view similarity, eliminating duplicate projections and residual background that temporal pruning alone cannot suppress. These two modules together constitute a complete spatio-temporal pruning process, preserving key scene information under strict compression. Validated across four benchmarks spanning perception, prediction, and planning, ST-Prune establishes new state-of-the-art for training-free token pruning. Notably, even at 90% token reduction, ST-Prune achieves near-lossless performance with certain metrics surpassing the full-model baseline, while maintaining inference speeds comparable to existing pruning approaches.

Scalable Vector Graphics (SVGs) function both as visual images and as structured code that encode rich geometric and layout information, yet most methods rasterize them and discard this symbolic organization. At the same time, recent sentence embedding methods produce strong text representations but do not naturally extend to visual or structured modalities. We propose a training-free, instruction-guided multimodal embedding framework that uses a Multimodal Large Language Model (MLLM) to map text, raster images, and SVG code into an aligned embedding space. We control the direction of embeddings through modality-specific instructions and structural SVG cues, eliminating the need for learned projection heads or contrastive training. Our method has two key components: (1) Multimodal Explicit One-word Limitation (mEOL), which instructs the MLLM to summarize any multimodal input into a single token whose hidden state serves as a compact semantic embedding. (2) A semantic SVG rewriting module that assigns meaningful identifiers and simplifies nested SVG elements through visual reasoning over the rendered image, exposing geometric and relational cues hidden in raw code. Using a repurposed VGBench, we build the first text-to-SVG retrieval benchmark and show that our training-free embeddings outperform encoder-based and training-based multimodal baselines. These results highlight prompt-level control as an effective alternative to parameter-level training for structure-aware multimodal retrieval. Project page: https://scene-the-ella.github.io/meol/

We introduce C-GenReg, a training-free framework for 3D point cloud registration that leverages the complementary strengths of world-scale generative priors and registration-oriented Vision Foundation Models (VFMs). Current learning-based 3D point cloud registration methods struggle to generalize across sensing modalities, sampling differences, and environments. Hence, C-GenReg augments the geometric point cloud registration branch by transferring the matching problem into an auxiliary image domain, where VFMs excel, using a World Foundation Model to synthesize multi-view-consistent RGB representations from the input geometry. This generative transfer, preserves spatial coherence across source and target views without any fine-tuning. From these generated views, a VFM pretrained for finding dense correspondences extracts matches. The resulting pixel correspondences are lifted back to 3D via the original depth maps. To further enhance robustness, we introduce a "Match-then-Fuse" probabilistic cold-fusion scheme that combines two independent correspondence posteriors, that of the generated-RGB branch with that of the raw geometric branch. This principled fusion preserves each modality inductive bias and provides calibrated confidence without any additional learning. C-GenReg is zero-shot and plug-and-play: all modules are pretrained and operate without fine-tuning. Extensive experiments on indoor (3DMatch, ScanNet) and outdoor (Waymo) benchmarks demonstrate strong zero-shot performance and superior cross-domain generalization. For the first time, we demonstrate a generative registration framework that operates successfully on real outdoor LiDAR data, where no imagery data is available.

As Large Language Models (LLMs) receive increasing attention and are being deployed across various domains, their potential risks, including generating harmful or biased content, producing unsupported claims, and exhibiting vulnerabilities to adversarial attacks, have drawn significant attention. To enable quick and low-cost adaptation, training-free methods have recently emerged as cost-effective alternatives to post-training alignment techniques. Despite their promising results, these methods are evaluated inconsistently across the literature, cover limited dimensions of trustworthiness, and can introduce undesirable side effects, such as utility degradation and increased brittleness. To fully assess the impacts of these training-free methods, we take a step back and systematically re-evaluate the effectiveness of existing training-free methods against various trustworthy settings and their influence on utility, robustness, and computational overhead. We also categorize these methods into three levels (input, internal, and output) based on where they intervene in the model's information flow during inference. Using this taxonomy, we conduct a comprehensive analysis of various representative and effective methods from each level across different LLM families and sizes. Our analysis highlights several trade-offs and unresolved challenges in current approaches. We summarize key findings and limitations in the existing literature, and propose practical recommendations for balancing trustworthiness, utility, and robustness in LLMs without the need for additional training.

Large audio-language models (LALMs) generalize across speech, sound, and music, but unified decoders can exhibit a \emph{temporal smoothing bias}: transient acoustic cues may be underutilized in favor of temporally smooth context that is better supported by language priors, leading to less specific audio-grounded outputs. We propose \emph{Temporal Contrastive Decoding} (TCD), a training-free decoding method for unified LALMs that mitigates this effect at inference time. TCD constructs a temporally blurred slow-path view by smoothing the input waveform and re-encoding it, then contrasts next-token logits from the original and slow-path views. The contrastive signal is applied as a token-level logit update restricted to a small candidate set. A self-normalized stability score sets the blur window and update scale, and a step-wise gate based on uncertainty and audio reliance activates the update only when needed. Experiments on MMAU and AIR-Bench show consistent improvements on strong unified LALMs. We further conduct ablations and an architectural applicability study to analyze the contributions of key components and how TCD behaves across large audio-language model designs.

Weakly supervised semantic segmentation aims to achieve pixel-level predictions using image-level labels. Existing methods typically entangle semantic recognition and object localization, which often leads models to focus exclusively on sparse discriminative regions. Although foundation models show immense potential, many approaches still follow the tightly coupled optimization paradigm, struggling to effectively alleviate pseudo-label noise and often relying on time-consuming multi-stage retraining or unstable end-to-end joint optimization. To address the above challenges, we present ModuSeg, a training-free weakly supervised semantic segmentation framework centered on explicitly decoupling object discovery and semantic assignment. Specifically, we integrate a general mask proposer to extract geometric proposals with reliable boundaries, while leveraging semantic foundation models to construct an offline feature bank, transforming segmentation into a non-parametric feature retrieval process. Furthermore, we propose semantic boundary purification and soft-masked feature aggregation strategies to effectively mitigate boundary ambiguity and quantization errors, thereby extracting high-quality category prototypes. Extensive experiments demonstrate that the proposed decoupled architecture better preserves fine boundaries without parameter fine-tuning and achieves highly competitive performance on standard benchmark datasets. Code is available at https://github.com/Autumnair007/ModuSeg.

Drone-view geo-localization (DVGL) aims to determine the location of drones in GPS-denied environments by retrieving the corresponding geotagged satellite tile from a reference gallery given UAV observations of a location. In many existing formulations, these observations are represented by a single oblique UAV image. In contrast, our satellite-free setting is designed for multi-view UAV sequences, which are used to construct a geometry-normalized UAV-side location representation before cross-view retrieval. Existing approaches rely on satellite imagery during training, either through paired supervision or unsupervised alignment, which limits practical deployment when satellite data are unavailable or restricted. In this paper, we propose a satellite-free training (SFT) framework that converts drone imagery into cross-view compatible representations through three main stages: drone-side 3D scene reconstruction, geometry-based pseudo-orthophoto generation, and satellite-free feature aggregation for retrieval. Specifically, we first reconstruct dense 3D scenes from multi-view drone images using 3D Gaussian splatting and project the reconstructed geometry into pseudo-orthophotos via PCA-guided orthographic projection. This rendering stage operates directly on reconstructed scene geometry without requiring camera parameters at rendering time. Next, we refine these orthophotos with lightweight geometry-guided inpainting to obtain texture-complete drone-side views. Finally, we extract DINOv3 patch features from the generated orthophotos, learn a Fisher vector aggregation model solely from drone data, and reuse it at test time to encode satellite tiles for cross-view retrieval. Experimental results on University-1652 and SUES-200 show that our SFT framework substantially outperforms satellite-free generalization baselines and narrows the gap to methods trained with satellite imagery.

Text-to-image generation powered by Diffusion Transformers (DiTs) has made remarkable strides, yet remote sensing (RS) synthesis lags behind due to two barriers: the absence of a domain-specialized DiT prior and the prohibitive cost of training at the large resolutions that RS applications demand. Training-free resolution promotion via Rotary Position Embedding (RoPE) rescaling offers a practical remedy, but every existing method applies a static positional scaling rule throughout the denoising process. This uniform compression is particularly harmful for RS imagery, whose substantially denser medium- and high-frequency energy encodes the fine structures critical for aerial-scene realism, such as vehicles, building contours, and road markings. Addressing both challenges requires a domain-specialized generative prior coupled with a denoising-aware positional adaptation strategy. To this end, we fine-tune FLUX on over 100,000 curated RS images to build a strong domain prior (RS-FLUX), and propose Spectrum-aware Highly-dynamic Adaptation for Resolution Promotion (SHARP), a training-free method that introduces a rational fractional time schedule k_rs(t) into RoPE. SHARP applies strong positional promotion during the early layout-formation stage and progressively relaxes it during detail recovery, aligning extrapolation strength with the frequency-progressive nature of diffusion denoising. Its resolution-agnostic formulation further enables robust multi-scale generation from a single set of hyperparameters. Extensive experiments across six square and rectangular resolutions show that SHARP consistently outperforms all training-free baselines on CLIP Score, Aesthetic Score, and HPSv2, with widening margins at more aggressive extrapolation factors and negligible computational overhead. Code and weights are available at https://github.com/bxuanz/SHARP.

Cross-View Geo-Localization (CVGL) in remote sensing aims to locate a drone-view query by matching it to geo-tagged satellite images. Although supervised methods have achieved strong results on close-set benchmarks, they often fail to generalize to unconstrained, real-world scenarios due to severe viewpoint differences and dataset bias. To overcome these limitations, we present VFM-Loc, a training-free CVGL framework that leverages the generalizable visual representations from vision foundational models (VFMs). VFM-Loc identifies and matches discriminative visual clues across different viewpoints through a progressive alignment strategy. First, we design a hierarchical clue extraction mechanism using Generalized Mean pooling and Scale-Weighted R-MAC to preserve distinctive visual clues across scales while maintaining hierarchical confidence. Second, we introduce a statistical manifold alignment pipeline based on domain-wise PCA and Orthogonal Procrustes analysis, linearly aligning heterogeneous feature distributions in a shared metric space. Experiments demonstrate that VFM-Loc exhibits high accuracy on standard benchmarks and surpasses supervised methods by over 20% in Recall@1 on the challenging LO-UCV dataset with large oblique angles. This work highlights that principled alignment of pre-trained features can effectively bridge the cross-view gap, establishing a robust and training-free paradigm for real-world CVGL. The relevant code is made available at: github.com/DingLei14/VFM-Loc.

While specialized Medical Vision-Language Models (VLMs) have achieved remarkable success in interpreting 2D and 3D medical modalities, their deployment for 3D volumetric data remains constrained by significant computational inefficiencies. Current architectures typically suffer from massive anatomical redundancy due to the direct concatenation of consecutive 2D slices and lack the flexibility to handle heterogeneous information densities across different slices using fixed pruning ratios. To address these challenges, we propose MedPruner, a training-free and model-agnostic hierarchical token pruning framework specifically designed for efficient 3D medical image understanding. MedPruner introduces a two-stage mechanism: an Inter-slice Anchor-based Filtering module to eliminate slice-level temporal redundancy, followed by a Dynamic Information Nucleus Selection strategy that achieves adaptive token-level compression by quantifying cumulative attention weights. Extensive experiments on three 3D medical benchmarks and across three diverse medical VLMs reveal massive token redundancy in existing architectures. Notably, MedPruner enables models such as MedGemma-1.5 to maintain or even exceed their original performance while retaining fewer than 5% of visual tokens, thereby reducing visual-token overhead and validating the necessity of dynamic token selection for practical clinical deployment. Our code is available at https://github.com/CUHK-AIM-Group/MedPruner.

Video Temporal Grounding (VTG) localizes the temporal boundaries of query-relevant moments in long, untrimmed videos, making video-language-model prohibitively expensive. While recent training-free token pruning has shown success in video question answering, naively applying these objectives to VTG causes drastic degradation, as VTG crucially depends on boundary-sensitive evidence and cross-frame reasoning chains. We therefore identify two VTG-specific pruning principles: evidence retention, which keeps query-critical patches especially around event boundaries, and connectivity strength, which preserves cross-frame connectivity for long-range evidence aggregation. Building on these insights, we propose SemVID, a training-free pruning framework that constructs a compact yet coherent token subset with complementary semantic roles. SemVID first allocates per-frame budgets by balancing query relevance and inter-frame variation to avoid over-pruned segments, and then selects three types of tokens: object tokens for diverse query-critical evidence, motion tokens to capture meaningful transitions and serve as cross-frame relays, and context tokens for scene continuity. Extensive experiments show that SemVID achieves a strong accuracy-efficiency trade-off, retaining up to 95.4% mIoU with only 12.5% visual tokens and delivering up to a 5.8x prefill speedup, consistently outperforming prior methods under the same budgets. Our code is available at https://github.com/JiaqiLi404/SemVID

Volume denoising is a foundational problem in computational imaging, as many 3D imaging inverse problems face high levels of measurement noise. Inspired by the strong 2D image denoising properties of Field of Junctions (ICCV 2021), we propose a novel, fully volumetric 3D Field of Junctions (3D FoJ) representation that optimizes a junction of 3D wedges that best explain each 3D patch of a full volume, while encouraging consistency between overlapping patches. In addition to direct volume denoising, we leverage our 3D FoJ representation as a structural prior that: (i) requires no training data, and thus precludes the risk of hallucination, (ii) preserves and enhances sharp edge and corner structures in 3D, even under low signal to noise ratio (SNR), and (iii) can be used as a drop-in denoising representation via projected or proximal gradient descent for any volumetric inverse problem with low SNR. We demonstrate successful volume reconstruction and denoising with 3D FoJ across three diverse 3D imaging tasks with low-SNR measurements: low-dose X-ray computed tomography (CT), cryogenic electron tomography (cryo-ET), and denoising point clouds such as those from lidar in adverse weather. Across these challenging low-SNR volumetric imaging problems, 3D FoJ outperforms the evaluated classical denoisers, untrained neural denoisers, and denoisers trained only on noisy examples. Code is available at https://github.com/voilalab/3D-Field-of-Junctions.

Image diversity remains a fundamental challenge for text-to-image diffusion models. Low-diversity generation often leads to repetitive outputs, increasing sampling redundancy and hindering both creative exploration and downstream applications. A key factor is the tendency of diffusion models to collapse toward strong modes in the learned distribution. Existing attempts to improve diversity, such as steering-based guidance, often introduce distortions that degrade image quality. To address this issue, we propose Token-Prompt Embedding Space Optimization (TPSO), a training-free and model-agnostic module. TPSO introduces learnable parameters to explore underrepresented regions of the token embedding space, reducing the tendency to repeatedly sample from strong modes of the distribution. Meanwhile, a prompt-level semantic constraint regulates distribution shifts, preventing quality degradation while preserving semantic fidelity. Extensive experiments on MS-COCO across three representative diffusion backbones demonstrate that TPSO substantially improves diversity, boosting performance from 1.10 to 4.18, while maintaining image quality with only a modest inference-time overhead of 3.6% to 8.9%. Code is available at: https://github.com/Open-Debin/TPSO.

Flow matching models generate high-fidelity molecular geometries but incur significant computational costs during inference, requiring hundreds of network evaluations. This inference overhead becomes the primary bottleneck when such models are employed in practice to sample large numbers of molecular candidates. This work discusses a training-free caching strategy that accelerates molecular geometry generation by predicting intermediate hidden states across solver steps. The proposed method operates directly on the SE(3)-equivariant backbone, is compatible with pretrained models, and is orthogonal to existing training-based accelerations and system-level optimizations. Experiments on the GEOM-Drugs dataset demonstrate that caching achieves a twofold reduction in wall-clock inference time at matched sample quality and a speedup of up to 3x compared to the base model with minimal sample quality degradation. Because these gains compound with other optimizations, applying caching alongside other general, lossless optimizations yield as much as a 7x speedup.

While video generation models excel at producing high-quality monocular videos, generating 3D stereoscopic and spatial videos for immersive applications remains an underexplored challenge. We present a pose-free and training-free method that leverages an off-the-shelf monocular video generation model to produce immersive 3D videos. Our approach first warps the generated monocular video into pre-defined camera viewpoints using estimated depth information, then applies a novel \textit{frame matrix} inpainting framework. This framework utilizes the original video generation model to synthesize missing content across different viewpoints and timestamps, ensuring spatial and temporal consistency without requiring additional model fine-tuning. Moreover, we develop a \dualupdate~scheme that further improves the quality of video inpainting by alleviating the negative effects propagated from disoccluded areas in the latent space. The resulting multi-view videos are then adapted into stereoscopic pairs or optimized into 4D Gaussians for spatial video synthesis. We validate the efficacy of our proposed method by conducting experiments on videos from various generative models, such as Sora, Lumiere, WALT, and Zeroscope. The experiments demonstrate that our method has a significant improvement over previous methods. Project page at: https://daipengwa.github.io/S-2VG_ProjectPage/

Recent advances in text-to-image (T2I) generation have led to impressive visual results. However, these models still face significant challenges when handling complex prompt, particularly those involving multiple subjects with distinct attributes. Inspired by the human drawing process, which first outlines the composition and then incrementally adds details, we propose Detail++, a training-free framework that introduces a novel Progressive Detail Injection (PDI) strategy to address this limitation. Specifically, we decompose a complex prompt into a sequence of simplified sub-prompts, guiding the generation process in stages. This staged generation leverages the inherent layout-controlling capacity of self-attention to first ensure global composition, followed by precise refinement. To achieve accurate binding between attributes and corresponding subjects, we exploit cross-attention mechanisms and further introduce a Centroid Alignment Loss at test time to reduce binding noise and enhance attribute consistency. Extensive experiments on T2I-CompBench and a newly constructed style composition benchmark demonstrate that Detail++ significantly outperforms existing methods, particularly in scenarios involving multiple objects and complex stylistic conditions.

Although large language models (LLMs) have achieved remarkable performance, the inherent stochasticity of their reasoning processes and varying conclusions present significant challenges. Majority voting or Best-of-N with external verifiers has been explored to mitigate this, but these approaches are limited in applicability or require additional training. To address this problem, we propose a novel framework that Recycles Few-shot examples to verify LLM outputs (ReFeri). Our key idea is to utilize the given few-shot examples not only to generate outputs, but also to evaluate the candidate outputs. Specifically, ReFeri combines a forward confidence score with a backward reconstruction penalty to select candidates that follow few-shot guidance while avoiding demonstration-specific overfitting. Experiments with three different LLMs across seven diverse tasks demonstrate that our framework significantly improves the accuracy of LLMs---achieving an average relative gain of 8.2%---through effective response selection.

Diffusion models have recently unlocked new possibilities in editing images of real-world objects. Yet, transforming objects in non-rigid ways, such as modifying poses or applying image-based conditioning, continues to present significant challenges. Retaining the unique identity of objects during these edits is a complex task, and current techniques often fall short of delivering the precision needed for industrial settings, where consistency is non-negotiable. Additionally, adapting diffusion models demands custom training data, which is often unavailable in real-world scenarios. To address these gaps, we present FashionRepose, a novel, training-free pipeline designed to handle non-rigid pose adjustments specifically for the fashion industry. This approach combines pretrained off-the-shelf models to modify the poses of long-sleeve garments while safeguarding their identity and branding characteristics. By adopting a zero-shot methodology, FashionRepose enables near real-time edits, entirely eliminating the requirement for specialized training data. FashionRepose has been deployed for a global fashion firm, OVS, handling more than 30,000 long-sleeve garments.