SDAR-VL: Stable and Efficient Block-wise Diffusion for Vision-Language Understanding

Block-wise discrete diffusion offers an attractive balance between parallel generation and causal dependency modeling, making it a promising backbone for vision-language modeling. However, its practical adoption has been limited by high training cost, slow convergence, and instability, which have so far kept it behind strong autoregressive (AR) baselines. We present \textbf{SDAR-VL}, the first systematic application of block-wise discrete diffusion to large-scale vision-language understanding (VLU), together with an \emph{integrated framework for efficient and stable training}. This framework unifies three components: (1) \textbf{Asynchronous Block-wise Noise Scheduling} to diversify supervision within each batch; (2) \textbf{Effective Mask Ratio Scaling} for unbiased loss normalization under stochastic masking; and (3) a \textbf{Progressive Beta Noise Curriculum} that increases effective mask coverage while preserving corruption diversity. Experiments on 21 single-image, multi-image, and video benchmarks show that SDAR-VL consistently improves \emph{training efficiency}, \emph{convergence stability}, and \emph{task performance} over conventional block diffusion. On this evaluation suite, SDAR-VL sets a new state of the art among diffusion-based vision-language models and, under matched settings, matches or surpasses strong AR baselines such as LLaVA-OneVision as well as the global diffusion baseline LLaDA-V, establishing block-wise diffusion as a practical backbone for VLU.

AsyncVLA: Asynchronous Flow Matching for Vision-Language-Action Models

Vision-language-action (VLA) models have recently emerged as a powerful paradigm for building generalist robots. However, traditional VLA models that generate actions through flow matching (FM) typically rely on rigid and uniform time schedules, i.e., synchronous FM (SFM). Without action context awareness and asynchronous self-correction, SFM becomes unstable in long-horizon tasks, where a single action error can cascade into failure. In this work, we propose asynchronous flow matching VLA (AsyncVLA), a novel framework that introduces temporal flexibility in asynchronous FM (AFM) and enables self-correction in action generation. AsyncVLA breaks from the vanilla SFM in VLA models by generating the action tokens in a non-uniform time schedule with action context awareness. Besides, our method introduces the confidence rater to extract confidence of the initially generated actions, enabling the model to selectively refine inaccurate action tokens before execution. Moreover, we propose a unified training procedure for SFM and AFM that endows a single model with both modes, improving KV-cache utilization. Extensive experiments on robotic manipulation benchmarks demonstrate that AsyncVLA is data-efficient and exhibits self-correction ability. AsyncVLA achieves state-of-the-art results across general embodied evaluations due to its asynchronous generation in AFM. Our code is available at https://github.com/YuhuaJiang2002/AsyncVLA.

Nirvana: A Specialized Generalist Model With Task-Aware Memory Mechanism

Specialized Generalist Models (SGMs) aim to preserve broad capabilities while achieving expert-level performance in target domains. However, traditional LLM structures including Transformer, Linear Attention, and hybrid models do not employ specialized memory mechanism guided by task information. In this paper, we present Nirvana, an SGM with specialized memory mechanism, linear time complexity, and test-time task information extraction. Besides, we propose the Task-Aware Memory Trigger ($\textit{Trigger}$) that flexibly adjusts memory mechanism based on the current task's requirements. In Trigger, each incoming sample is treated as a self-supervised fine-tuning task, enabling Nirvana to adapt its task-related parameters on the fly to domain shifts. We also design the Specialized Memory Updater ($\textit{Updater}$) that dynamically memorizes the context guided by Trigger. We conduct experiments on both general language tasks and specialized medical tasks. On a variety of natural language modeling benchmarks, Nirvana achieves competitive or superior results compared to the existing LLM structures. To prove the effectiveness of Trigger on specialized tasks, we test Nirvana's performance on a challenging medical task, i.e., Magnetic Resonance Imaging (MRI). We post-train frozen Nirvana backbone with lightweight codecs on paired electromagnetic signals and MRI images. Despite the frozen Nirvana backbone, Trigger guides the model to adapt to the MRI domain with the change of task-related parameters. Nirvana achieves higher-quality MRI reconstruction compared to conventional MRI models as well as the models with traditional LLMs' backbone, and can also generate accurate preliminary clinical reports accordingly.

DPMT: Dual Process Multi-scale Theory of Mind Framework for Real-time Human-AI Collaboration

Real-time human-artificial intelligence (AI) collaboration is crucial yet challenging, especially when AI agents must adapt to diverse and unseen human behaviors in dynamic scenarios. Existing large language model (LLM) agents often fail to accurately model the complex human mental characteristics such as domain intentions, especially in the absence of direct communication. To address this limitation, we propose a novel dual process multi-scale theory of mind (DPMT) framework, drawing inspiration from cognitive science dual process theory. Our DPMT framework incorporates a multi-scale theory of mind (ToM) module to facilitate robust human partner modeling through mental characteristic reasoning. Experimental results demonstrate that DPMT significantly enhances human-AI collaboration, and ablation studies further validate the contributions of our multi-scale ToM in the slow system.

PAG: Multi-Turn Reinforced LLM Self-Correction with Policy as Generative Verifier

Large Language Models (LLMs) have demonstrated impressive capabilities in complex reasoning tasks, yet they still struggle to reliably verify the correctness of their own outputs. Existing solutions to this verification challenge often depend on separate verifier models or require multi-stage self-correction training pipelines, which limit scalability. In this paper, we propose Policy as Generative Verifier (PAG), a simple and effective framework that empowers LLMs to self-correct by alternating between policy and verifier roles within a unified multi-turn reinforcement learning (RL) paradigm. Distinct from prior approaches that always generate a second attempt regardless of model confidence, PAG introduces a selective revision mechanism: the model revises its answer only when its own generative verification step detects an error. This verify-then-revise workflow not only alleviates model collapse but also jointly enhances both reasoning and verification abilities. Extensive experiments across diverse reasoning benchmarks highlight PAG's dual advancements: as a policy, it enhances direct generation and self-correction accuracy; as a verifier, its self-verification outperforms self-consistency.

Fewer May Be Better: Enhancing Offline Reinforcement Learning with Reduced Dataset

Offline reinforcement learning (RL) represents a significant shift in RL research, allowing agents to learn from pre-collected datasets without further interaction with the environment. A key, yet underexplored, challenge in offline RL is selecting an optimal subset of the offline dataset that enhances both algorithm performance and training efficiency. Reducing dataset size can also reveal the minimal data requirements necessary for solving similar problems. In response to this challenge, we introduce ReDOR (Reduced Datasets for Offline RL), a method that frames dataset selection as a gradient approximation optimization problem. We demonstrate that the widely used actor-critic framework in RL can be reformulated as a submodular optimization objective, enabling efficient subset selection. To achieve this, we adapt orthogonal matching pursuit (OMP), incorporating several novel modifications tailored for offline RL. Our experimental results show that the data subsets identified by ReDOR not only boost algorithm performance but also do so with significantly lower computational complexity.

Episodic Novelty Through Temporal Distance

Exploration in sparse reward environments remains a significant challenge in reinforcement learning, particularly in Contextual Markov Decision Processes (CMDPs), where environments differ across episodes. Existing episodic intrinsic motivation methods for CMDPs primarily rely on count-based approaches, which are ineffective in large state spaces, or on similarity-based methods that lack appropriate metrics for state comparison. To address these shortcomings, we propose Episodic Novelty Through Temporal Distance (ETD), a novel approach that introduces temporal distance as a robust metric for state similarity and intrinsic reward computation. By employing contrastive learning, ETD accurately estimates temporal distances and derives intrinsic rewards based on the novelty of states within the current episode. Extensive experiments on various benchmark tasks demonstrate that ETD significantly outperforms state-of-the-art methods, highlighting its effectiveness in enhancing exploration in sparse reward CMDPs.

Electromagnetic Property Sensing and Channel Reconstruction Based on Diffusion Schrödinger Bridge in ISAC

Integrated sensing and communications (ISAC) has emerged as a transformative paradigm for next-generation wireless systems. In this paper, we present a novel ISAC scheme that leverages the diffusion Schrodinger bridge (DSB) to realize the sensing of electromagnetic (EM) property of a target as well as the reconstruction of the wireless channel. The DSB framework connects EM property sensing and channel reconstruction by establishing a bidirectional process: the forward process transforms the distribution of EM property into the channel distribution, while the reverse process reconstructs the EM property from the channel. To handle the difference in dimensionality between the high-dimensional sensing channel and the lower-dimensional EM property, we generate latent representations using an autoencoder network. The autoencoder compresses the sensing channel into a latent space that retains essential features, which incorporates positional embeddings to process spatial context. The simulation results demonstrate the effectiveness of the proposed DSB framework, which achieves superior reconstruction of the targets shape, relative permittivity, and conductivity. Moreover, the proposed method can also realize high-fidelity channel reconstruction given the EM property of the target. The dual capability of accurately sensing the EM property and reconstructing the channel across various positions within the sensing area underscores the versatility and potential of the proposed approach for broad application in future ISAC systems.

Electromagnetic Property Sensing Based on Diffusion Model in ISAC System

Integrated sensing and communications (ISAC) has opened up numerous game-changing opportunities for future wireless systems. In this paper, we develop a novel ISAC scheme that utilizes the diffusion model to sense the electromagnetic (EM) property of the target in a predetermined sensing area. Specifically, we first estimate the sensing channel by using both the communications and the sensing signals echoed back from the target. Then we employ the diffusion model to generate the point cloud that represents the target and thus enables 3D visualization of the target's EM property distribution. In order to minimize the mean Chamfer distance (MCD) between the ground truth and the estimated point clouds, we further design the communications and sensing beamforming matrices under the constraint of a maximum transmit power and a minimum communications achievable rate for each user equipment (UE). Simulation results demonstrate the efficacy of the proposed method in achieving high-quality reconstruction of the target's shape, relative permittivity, and conductivity. Besides, the proposed method can sense the EM property of the target effectively in any position of the sensing area.

Integrated Sensing and Communications Framework for 6G Networks

In this paper, we propose a novel integrated sensing and communications (ISAC) framework for the sixth generation (6G) mobile networks, in which we decompose the real physical world into static environment, dynamic targets, and various object materials. The ubiquitous static environment occupies the vast majority of the physical world, for which we design static environment reconstruction (SER) scheme to obtain the layout and point cloud information of static buildings. The dynamic targets floating in static environments create the spatiotemporal transition of the physical world, for which we design comprehensive dynamic target sensing (DTS) scheme to detect, estimate, track, image and recognize the dynamic targets in real-time. The object materials enrich the electromagnetic laws of the physical world, for which we develop object material recognition (OMR) scheme to estimate the electromagnetic coefficient of the objects. Besides, to integrate these sensing functions into existing communications systems, we discuss the interference issues and corresponding solutions for ISAC cellular networks. Furthermore, we develop an ISAC hardware prototype platform that can reconstruct the environmental maps and sense the dynamic targets while maintaining communications services. With all these designs, the proposed ISAC framework can support multifarious emerging applications, such as digital twins, low altitude economy, internet of vehicles, marine management, deformation monitoring, etc.