Functionality-Oriented LLM Merging on the Fisher--Rao Manifold

Weight-space merging aims to combine multiple fine-tuned LLMs into a single model without retraining, yet most existing approaches remain fundamentally parameter-space heuristics. This creates three practical limitations. First, linear averaging, task vectors, and related rules operate on Euclidean coordinates, even though the desired goal is to merge functionality, i.e., predictive behaviors across tasks. Second, when the source checkpoints are farther apart or more heterogeneous, Euclidean blends often trigger representation collapse, manifested as activation variance shrinkage and effective-rank degradation, which sharply degrades accuracy. Third, many geometry-inspired methods are most natural for two-model interpolation and do not extend cleanly to merging N>2 experts with a principled objective. We address these issues by formulating model merging as computing a weighted Karcher mean on the Fisher--Rao manifold, which is locally equivalent to minimizing a KL-based function distance between predictive distributions. We derive a practical fixed-point algorithm using a lightweight spherical proxy that preserves norms and generalizes directly to multi-expert merging. Across various benchmarks and collapse diagnostics, our method remains stable as the number and heterogeneity of merged models increase, consistently outperforming prior baselines.

PreciseCache: Precise Feature Caching for Efficient and High-fidelity Video Generation

High computational costs and slow inference hinder the practical application of video generation models. While prior works accelerate the generation process through feature caching, they often suffer from notable quality degradation. In this work, we reveal that this issue arises from their inability to distinguish truly redundant features, which leads to the unintended skipping of computations on important features. To address this, we propose \textbf{PreciseCache}, a plug-and-play framework that precisely detects and skips truly redundant computations, thereby accelerating inference without sacrificing quality. Specifically, PreciseCache contains two components: LFCache for step-wise caching and BlockCache for block-wise caching. For LFCache, we compute the Low-Frequency Difference (LFD) between the prediction features of the current step and those from the previous cached step. Empirically, we observe that LFD serves as an effective measure of step-wise redundancy, accurately detecting highly redundant steps whose computation can be skipped through reusing cached features. To further accelerate generation within each non-skipped step, we propose BlockCache, which precisely detects and skips redundant computations at the block level within the network. Extensive experiments on various backbones demonstrate the effectiveness of our PreciseCache, such as achieving an average of $2.6\times$ speedup on Wan2.1-14B without noticeable quality loss.

SkillOrchestra: Learning to Route Agents via Skill Transfer

Compound AI systems promise capabilities beyond those of individual models, yet their success depends critically on effective orchestration. Existing routing approaches face two limitations: (1) input-level routers make coarse query-level decisions that ignore evolving task requirements; (2) RL-trained orchestrators are expensive to adapt and often suffer from routing collapse, repeatedly invoking one strong but costly option in multi-turn scenarios. We introduce SkillOrchestra, a framework for skill-aware orchestration. Instead of directly learning a routing policy end-to-end, SkillOrchestra learns fine-grained skills from execution experience and models agent-specific competence and cost under those skills. At deployment, the orchestrator infers the skill demands of the current interaction and selects agents that best satisfy them under an explicit performance-cost trade-off. Extensive experiments across ten benchmarks demonstrate that SkillOrchestra outperforms SoTA RL-based orchestrators by up to 22.5% with 700x and 300x learning cost reduction compared to Router-R1 and ToolOrchestra, respectively. These results show that explicit skill modeling enables scalable, interpretable, and sample-efficient orchestration, offering a principled alternative to data-intensive RL-based approaches. The code is available at: https://github.com/jiayuww/SkillOrchestra.

SPD-Faith Bench: Diagnosing and Improving Faithfulness in Chain-of-Thought for Multimodal Large Language Models

Chain-of-Thought reasoning is widely used to improve the interpretability of multimodal large language models (MLLMs), yet the faithfulness of the generated reasoning traces remains unclear. Prior work has mainly focused on perceptual hallucinations, leaving reasoning level unfaithfulness underexplored. To isolate faithfulness from linguistic priors, we introduce SPD-Faith Bench, a diagnostic benchmark based on fine-grained image difference reasoning that enforces explicit visual comparison. Evaluations on state-of-the-art MLLMs reveal two systematic failure modes, perceptual blindness and perception-reasoning dissociation. We trace these failures to decaying visual attention and representation shifts in the residual stream. Guided by this analysis, we propose SAGE, a train-free visual evidence-calibrated framework that improves visual routing and aligns reasoning with perception. Our results highlight the importance of explicitly evaluating faithfulness beyond response correctness. Our benchmark and codes are available at https://github.com/Johanson-colab/SPD-Faith-Bench.

LiveResearchBench: A Live Benchmark for User-Centric Deep Research in the Wild

Deep research -- producing comprehensive, citation-grounded reports by searching and synthesizing information from hundreds of live web sources -- marks an important frontier for agentic systems. To rigorously evaluate this ability, four principles are essential: tasks should be (1) user-centric, reflecting realistic information needs, (2) dynamic, requiring up-to-date information beyond parametric knowledge, (3) unambiguous, ensuring consistent interpretation across users, and (4) multi-faceted and search-intensive, requiring search over numerous web sources and in-depth analysis. Existing benchmarks fall short of these principles, often focusing on narrow domains or posing ambiguous questions that hinder fair comparison. Guided by these principles, we introduce LiveResearchBench, a benchmark of 100 expert-curated tasks spanning daily life, enterprise, and academia, each requiring extensive, dynamic, real-time web search and synthesis. Built with over 1,500 hours of human labor, LiveResearchBench provides a rigorous basis for systematic evaluation. To evaluate citation-grounded long-form reports, we introduce DeepEval, a comprehensive suite covering both content- and report-level quality, including coverage, presentation, citation accuracy and association, consistency and depth of analysis. DeepEval integrates four complementary evaluation protocols, each designed to ensure stable assessment and high agreement with human judgments. Using LiveResearchBench and DeepEval, we conduct a comprehensive evaluation of 17 frontier deep research systems, including single-agent web search, single-agent deep research, and multi-agent systems. Our analysis reveals current strengths, recurring failure modes, and key system components needed to advance reliable, insightful deep research.

Astra: Toward General-Purpose Mobile Robots via Hierarchical Multimodal Learning

Modern robot navigation systems encounter difficulties in diverse and complex indoor environments. Traditional approaches rely on multiple modules with small models or rule-based systems and thus lack adaptability to new environments. To address this, we developed Astra, a comprehensive dual-model architecture, Astra-Global and Astra-Local, for mobile robot navigation. Astra-Global, a multimodal LLM, processes vision and language inputs to perform self and goal localization using a hybrid topological-semantic graph as the global map, and outperforms traditional visual place recognition methods. Astra-Local, a multitask network, handles local path planning and odometry estimation. Its 4D spatial-temporal encoder, trained through self-supervised learning, generates robust 4D features for downstream tasks. The planning head utilizes flow matching and a novel masked ESDF loss to minimize collision risks for generating local trajectories, and the odometry head integrates multi-sensor inputs via a transformer encoder to predict the relative pose of the robot. Deployed on real in-house mobile robots, Astra achieves high end-to-end mission success rate across diverse indoor environments.

Beyond Accuracy: Dissecting Mathematical Reasoning for LLMs Under Reinforcement Learning

Reinforcement learning (RL) has become the dominant paradigm for endowing language models with advanced reasoning capabilities. Despite the substantial empirical gains demonstrated by RL-based training methods like GRPO, a granular understanding of their advantages is still lacking. To address this gap, we introduce a fine-grained analytic framework to dissect the impact of RL on reasoning. Our framework specifically investigates key elements that have been hypothesized to benefit from RL training: (1) plan-following and execution, (2) problem decomposition, and (3) improved reasoning and knowledge utilization. Using this framework, we gain insights beyond mere accuracy. For instance, providing models with explicit step-by-step plans surprisingly degrades performance on the most challenging benchmarks, yet RL-tuned models exhibit greater robustness, experiencing markedly smaller performance drops than their base counterparts. This suggests that RL may not primarily enhance the execution of external plans but rather empower models to formulate and follow internal strategies better suited to their reasoning processes. Conversely, we observe that RL enhances the model's capacity to integrate provided knowledge into its reasoning process, leading to performance improvements across diverse tasks. We also study difficulty, showing improved training by developing new ways to exploit hard problems. Our findings lay a foundation for more principled training and evaluation of reasoning models.

OmniGenBench: A Benchmark for Omnipotent Multimodal Generation across 50+ Tasks

Recent breakthroughs in large multimodal models (LMMs), such as the impressive GPT-4o-Native, have demonstrated remarkable proficiency in following general-purpose instructions for image generation. However, current benchmarks often lack the necessary breadth and depth to fully evaluate the diverse capabilities of these models. To overcome this limitation, we introduce OmniGenBench, a novel and comprehensive benchmark meticulously designed to assess the instruction-following abilities of state-of-the-art LMMs across both perception-centric and cognition-centric dimensions. Our OmniGenBench includes 57 diverse sub-tasks grounded in real-world scenarios, systematically categorized according to the specific model capabilities they demand. For rigorous evaluation, we further employ a dual-mode protocol. This protocol utilizes off-the-shelf visual parsing tools for perception-centric tasks and a powerful LLM-based judger for cognition-centric tasks to assess the alignment between generated images and user instructions. Using OmniGenBench, we evaluate mainstream generative models, including prevalent models like GPT-4o, Gemini-2.0-Flash, and Seedream, and provide in-depth comparisons and analyses of their performance.Code and data are available at https://github.com/emilia113/OmniGenBench.

COSMOS: Predictable and Cost-Effective Adaptation of LLMs

Large language models (LLMs) achieve remarkable performance across numerous tasks by using a diverse array of adaptation strategies. However, optimally selecting a model and adaptation strategy under resource constraints is challenging and often requires extensive experimentation. We investigate whether it is possible to accurately predict both performance and cost without expensive trials. We formalize the strategy selection problem for LLMs and introduce COSMOS, a unified prediction framework that efficiently estimates adaptation outcomes at minimal cost. We instantiate and study the capability of our framework via a pair of powerful predictors: embedding-augmented lightweight proxy models to predict fine-tuning performance, and low-sample scaling laws to forecast retrieval-augmented in-context learning. Extensive evaluation across eight representative benchmarks demonstrates that COSMOS achieves high prediction accuracy while reducing computational costs by 92.72% on average, and up to 98.71% in resource-intensive scenarios. Our results show that efficient prediction of adaptation outcomes is not only feasible but can substantially reduce the computational overhead of LLM deployment while maintaining performance standards.

NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results

This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images. This challenge received a wide range of impressive solutions, which are developed and evaluated using our collected real-world Raindrop Clarity dataset. Unlike existing deraining datasets, our Raindrop Clarity dataset is more diverse and challenging in degradation types and contents, which includes day raindrop-focused, day background-focused, night raindrop-focused, and night background-focused degradations. This dataset is divided into three subsets for competition: 14,139 images for training, 240 images for validation, and 731 images for testing. The primary objective of this challenge is to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions. There are a total of 361 participants in the competition, and 32 teams submitting valid solutions and fact sheets for the final testing phase. These submissions achieved state-of-the-art (SOTA) performance on the Raindrop Clarity dataset. The project can be found at https://lixinustc.github.io/CVPR-NTIRE2025-RainDrop-Competition.github.io/.