On the Generalization of SFT: A Reinforcement Learning Perspective with Reward Rectification

We present a simple yet theoretically motivated improvement to Supervised Fine-Tuning (SFT) for the Large Language Model (LLM), addressing its limited generalization compared to reinforcement learning (RL). Through mathematical analysis, we reveal that standard SFT gradients implicitly encode a problematic reward structure that may severely restrict the generalization capabilities of model. To rectify this, we propose Dynamic Fine-Tuning (DFT), stabilizing gradient updates for each token by dynamically rescaling the objective function with the probability of this token. Remarkably, this single-line code change significantly outperforms standard SFT across multiple challenging benchmarks and base models, demonstrating greatly improved generalization. Additionally, our approach shows competitive results in offline RL settings, offering an effective yet simpler alternative. This work bridges theoretical insight and practical solutions, substantially advancing SFT performance. The code will be available at https://github.com/yongliang-wu/DFT.

Learning Deblurring Texture Prior from Unpaired Data with Diffusion Model

Since acquiring large amounts of realistic blurry-sharp image pairs is difficult and expensive, learning blind image deblurring from unpaired data is a more practical and promising solution. Unfortunately, dominant approaches rely heavily on adversarial learning to bridge the gap from blurry domains to sharp domains, ignoring the complex and unpredictable nature of real-world blur patterns. In this paper, we propose a novel diffusion model (DM)-based framework, dubbed \ours, for image deblurring by learning spatially varying texture prior from unpaired data. In particular, \ours performs DM to generate the prior knowledge that aids in recovering the textures of blurry images. To implement this, we propose a Texture Prior Encoder (TPE) that introduces a memory mechanism to represent the image textures and provides supervision for DM training. To fully exploit the generated texture priors, we present the Texture Transfer Transformer layer (TTformer), in which a novel Filter-Modulated Multi-head Self-Attention (FM-MSA) efficiently removes spatially varying blurring through adaptive filtering. Furthermore, we implement a wavelet-based adversarial loss to preserve high-frequency texture details. Extensive evaluations show that \ours provides a promising unsupervised deblurring solution and outperforms SOTA methods in widely-used benchmarks.

Frequency Domain-Based Diffusion Model for Unpaired Image Dehazing

Unpaired image dehazing has attracted increasing attention due to its flexible data requirements during model training. Dominant methods based on contrastive learning not only introduce haze-unrelated content information, but also ignore haze-specific properties in the frequency domain (\ie,~haze-related degradation is mainly manifested in the amplitude spectrum). To address these issues, we propose a novel frequency domain-based diffusion model, named \ours, for fully exploiting the beneficial knowledge in unpaired clear data. In particular, inspired by the strong generative ability shown by Diffusion Models (DMs), we tackle the dehazing task from the perspective of frequency domain reconstruction and perform the DMs to yield the amplitude spectrum consistent with the distribution of clear images. To implement it, we propose an Amplitude Residual Encoder (ARE) to extract the amplitude residuals, which effectively compensates for the amplitude gap from the hazy to clear domains, as well as provide supervision for the DMs training. In addition, we propose a Phase Correction Module (PCM) to eliminate artifacts by further refining the phase spectrum during dehazing with a simple attention mechanism. Experimental results demonstrate that our \ours outperforms other state-of-the-art methods on both synthetic and real-world datasets.

HoliGS: Holistic Gaussian Splatting for Embodied View Synthesis

We propose HoliGS, a novel deformable Gaussian splatting framework that addresses embodied view synthesis from long monocular RGB videos. Unlike prior 4D Gaussian splatting and dynamic NeRF pipelines, which struggle with training overhead in minute-long captures, our method leverages invertible Gaussian Splatting deformation networks to reconstruct large-scale, dynamic environments accurately. Specifically, we decompose each scene into a static background plus time-varying objects, each represented by learned Gaussian primitives undergoing global rigid transformations, skeleton-driven articulation, and subtle non-rigid deformations via an invertible neural flow. This hierarchical warping strategy enables robust free-viewpoint novel-view rendering from various embodied camera trajectories by attaching Gaussians to a complete canonical foreground shape (\eg, egocentric or third-person follow), which may involve substantial viewpoint changes and interactions between multiple actors. Our experiments demonstrate that \ourmethod~ achieves superior reconstruction quality on challenging datasets while significantly reducing both training and rendering time compared to state-of-the-art monocular deformable NeRFs. These results highlight a practical and scalable solution for EVS in real-world scenarios. The source code will be released.

CoCo4D: Comprehensive and Complex 4D Scene Generation

Existing 4D synthesis methods primarily focus on object-level generation or dynamic scene synthesis with limited novel views, restricting their ability to generate multi-view consistent and immersive dynamic 4D scenes. To address these constraints, we propose a framework (dubbed as CoCo4D) for generating detailed dynamic 4D scenes from text prompts, with the option to include images. Our method leverages the crucial observation that articulated motion typically characterizes foreground objects, whereas background alterations are less pronounced. Consequently, CoCo4D divides 4D scene synthesis into two responsibilities: modeling the dynamic foreground and creating the evolving background, both directed by a reference motion sequence. Given a text prompt and an optional reference image, CoCo4D first generates an initial motion sequence utilizing video diffusion models. This motion sequence then guides the synthesis of both the dynamic foreground object and the background using a novel progressive outpainting scheme. To ensure seamless integration of the moving foreground object within the dynamic background, CoCo4D optimizes a parametric trajectory for the foreground, resulting in realistic and coherent blending. Extensive experiments show that CoCo4D achieves comparable or superior performance in 4D scene generation compared to existing methods, demonstrating its effectiveness and efficiency. More results are presented on our website https://colezwhy.github.io/coco4d/.

Dense360: Dense Understanding from Omnidirectional Panoramas

Multimodal Large Language Models (MLLMs) require comprehensive visual inputs to achieve dense understanding of the physical world. While existing MLLMs demonstrate impressive world understanding capabilities through limited field-of-view (FOV) visual inputs (e.g., 70 degree), we take the first step toward dense understanding from omnidirectional panoramas. We first introduce an omnidirectional panoramas dataset featuring a comprehensive suite of reliability-scored annotations. Specifically, our dataset contains 160K panoramas with 5M dense entity-level captions, 1M unique referring expressions, and 100K entity-grounded panoramic scene descriptions. Compared to multi-view alternatives, panoramas can provide more complete, compact, and continuous scene representations through equirectangular projections (ERP). However, the use of ERP introduces two key challenges for MLLMs: i) spatial continuity along the circle of latitude, and ii) latitude-dependent variation in information density. We address these challenges through ERP-RoPE, a position encoding scheme specifically designed for panoramic ERP. In addition, we introduce Dense360-Bench, the first benchmark for evaluating MLLMs on omnidirectional captioning and grounding, establishing a comprehensive framework for advancing dense visual-language understanding in panoramic settings.

CyberV: Cybernetics for Test-time Scaling in Video Understanding

Current Multimodal Large Language Models (MLLMs) may struggle with understanding long or complex videos due to computational demands at test time, lack of robustness, and limited accuracy, primarily stemming from their feed-forward processing nature. These limitations could be more severe for models with fewer parameters. To address these limitations, we propose a novel framework inspired by cybernetic principles, redesigning video MLLMs as adaptive systems capable of self-monitoring, self-correction, and dynamic resource allocation during inference. Our approach, CyberV, introduces a cybernetic loop consisting of an MLLM Inference System, a Sensor, and a Controller. Specifically, the sensor monitors forward processes of the MLLM and collects intermediate interpretations, such as attention drift, then the controller determines when and how to trigger self-correction and generate feedback to guide the next round. This test-time adaptive scaling framework enhances frozen MLLMs without requiring retraining or additional components. Experiments demonstrate significant improvements: CyberV boosts Qwen2.5-VL-7B by 8.3% and InternVL3-8B by 5.5% on VideoMMMU, surpassing the competitive proprietary model GPT-4o. When applied to Qwen2.5-VL-72B, it yields a 10.0% improvement, achieving performance even comparable to human experts. Furthermore, our method demonstrates consistent gains on general-purpose benchmarks, such as VideoMME and WorldSense, highlighting its effectiveness and generalization capabilities in making MLLMs more robust and accurate for dynamic video understanding. The code is released at https://github.com/marinero4972/CyberV.

RAPID Hand: A Robust, Affordable, Perception-Integrated, Dexterous Manipulation Platform for Generalist Robot Autonomy

This paper addresses the scarcity of low-cost but high-dexterity platforms for collecting real-world multi-fingered robot manipulation data towards generalist robot autonomy. To achieve it, we propose the RAPID Hand, a co-optimized hardware and software platform where the compact 20-DoF hand, robust whole-hand perception, and high-DoF teleoperation interface are jointly designed. Specifically, RAPID Hand adopts a compact and practical hand ontology and a hardware-level perception framework that stably integrates wrist-mounted vision, fingertip tactile sensing, and proprioception with sub-7 ms latency and spatial alignment. Collecting high-quality demonstrations on high-DoF hands is challenging, as existing teleoperation methods struggle with precision and stability on complex multi-fingered systems. We address this by co-optimizing hand design, perception integration, and teleoperation interface through a universal actuation scheme, custom perception electronics, and two retargeting constraints. We evaluate the platform's hardware, perception, and teleoperation interface. Training a diffusion policy on collected data shows superior performance over prior works, validating the system's capability for reliable, high-quality data collection. The platform is constructed from low-cost and off-the-shelf components and will be made public to ensure reproducibility and ease of adoption.

Conditional Panoramic Image Generation via Masked Autoregressive Modeling

Recent progress in panoramic image generation has underscored two critical limitations in existing approaches. First, most methods are built upon diffusion models, which are inherently ill-suited for equirectangular projection (ERP) panoramas due to the violation of the identically and independently distributed (i.i.d.) Gaussian noise assumption caused by their spherical mapping. Second, these methods often treat text-conditioned generation (text-to-panorama) and image-conditioned generation (panorama outpainting) as separate tasks, relying on distinct architectures and task-specific data. In this work, we propose a unified framework, Panoramic AutoRegressive model (PAR), which leverages masked autoregressive modeling to address these challenges. PAR avoids the i.i.d. assumption constraint and integrates text and image conditioning into a cohesive architecture, enabling seamless generation across tasks. To address the inherent discontinuity in existing generative models, we introduce circular padding to enhance spatial coherence and propose a consistency alignment strategy to improve generation quality. Extensive experiments demonstrate competitive performance in text-to-image generation and panorama outpainting tasks while showcasing promising scalability and generalization capabilities.

BusterX: MLLM-Powered AI-Generated Video Forgery Detection and Explanation

Advances in AI generative models facilitate super-realistic video synthesis, amplifying misinformation risks via social media and eroding trust in digital content. Several research works have explored new deepfake detection methods on AI-generated images to alleviate these risks. However, with the fast development of video generation models, such as Sora and WanX, there is currently a lack of large-scale, high-quality AI-generated video datasets for forgery detection. In addition, existing detection approaches predominantly treat the task as binary classification, lacking explainability in model decision-making and failing to provide actionable insights or guidance for the public. To address these challenges, we propose \textbf{GenBuster-200K}, a large-scale AI-generated video dataset featuring 200K high-resolution video clips, diverse latest generative techniques, and real-world scenes. We further introduce \textbf{BusterX}, a novel AI-generated video detection and explanation framework leveraging multimodal large language model (MLLM) and reinforcement learning for authenticity determination and explainable rationale. To our knowledge, GenBuster-200K is the {\it \textbf{first}} large-scale, high-quality AI-generated video dataset that incorporates the latest generative techniques for real-world scenarios. BusterX is the {\it \textbf{first}} framework to integrate MLLM with reinforcement learning for explainable AI-generated video detection. Extensive comparisons with state-of-the-art methods and ablation studies validate the effectiveness and generalizability of BusterX. The code, models, and datasets will be released.