Reinforcement Learning (RL) is crucial for empowering Video-LLMs with complex spatiotemporal reasoning.However, current RL paradigms predominantly rely on random data shuffling or naive curriculum strategies based on scalar difficulty metrics.We argue that scalar metrics fail to disentangle two orthogonal challenges in video understanding: Visual-Temporal Perception Load and Cognitive Reasoning Depth.To address this, we propose VideoCuRL, a novel framework that decomposes difficulty into these two axes.We employ efficient, training-free proxies-optical flow/keyframe entropy for visual complexity and Calibrated Surprisal for cognitive complexity-to map data onto a 2D curriculum grid.A competence-aware Diagonal Wavefront strategy then schedules training from base alignment to complex reasoning.Furthermore, we introduce Dynamic Sparse KL and Structured Revisiting to stabilize training against reward collapse and catastrophic forgetting.Extensive experiments show that VideoCuRL surpasses strong RL baselines on reasoning (+2.5% on VSI-Bench) and perception (+2.9% on VideoMME) tasks.Notably, VideoCuRL eliminates the prohibitive inference overhead of generation-based curricula, offering a scalable solution for robust video post-training.

This paper presents the design of a wideband pipelined multi-stage noise shaping (MASH) continuous time (CT) delta-sigma modulator (DSM). The quantization error of the overall$1^{\mathrm {st}}$-stage DSM is extracted as the input of the$2^{\mathrm {nd}}$stage, while the outputs of both stages are simply combined without using any digital filters. Overall, different shaping functions are generated for both QN without requiring any digital QN cancellation. Therefore, the pipelined MASH (PMASH) significantly mitigates QN leakage while retaining the decent loop stability of a traditional MASH. Additionally, several analyses have been made for the PMASH topology, e.g. the design guideline, the signal transfer function (STF), the robustness, etc. Clocked at 800MHz and enabling on-chip DAC calibration, the 65nm CMOS prototype with an exemplary 2-2 topology using multi-bit quantizers achieves 82.6 dB SNDR, 98.8 dB SFDR over 15.4 MHz BW at −0.5 dBFS 1.8 MHz input. The power consumption is 16.9 mW with 1.2V/1.5V supplies. It results in a competitive FoM${}_{\mathrm {S\vert SNDR}}$of 172.2 dB, while it avoids any off-chip calibrations.

The recently emerged movable antenna (MA) shows great potential in leveraging spatial degrees of freedom for enhancing the performance of wireless systems. However, resource allocation in MA-aided systems faces unique challenges due to the non-convex and coupled constraints on antenna positions. This paper systematically reveals the challenges brought by the minimum MA separation constraints, and proposes a penalty framework for resource allocation under such new constraints in MA-aided systems. By introducing auxiliary variables, the proposed framework separates the non-convex and coupled antenna distance constraints from the movable region constraint. This enables the resulting problem be efficiently solved by alternating optimization, where the optimization of the original variables resembles that in conventional resource allocation problem while the optimization with respect to the auxiliary variables is achieved in closed-form solutions. To illustrate the effectiveness of the proposed framework, we present three case studies: capacity maximization, latency minimization, and regularized zero-forcing precoding. Simulation results demonstrate that the proposed optimization framework consistently outperforms state-of-the-art schemes.

This paper addresses the challenge of large model (LM)-embedded wireless network for handling the trade-off problem of model accuracy and network latency. To guarantee a high-quality of users’ service, the network latency should be minimized while maintaining an acceptable inference accuracy. To meet this requirement, LM quantization is proposed to reduce the latency. However, the excessive quantization may destroy the accuracy of LM inference. To this end, a promising fluid antenna (FA) technology is investigated for enhancing the transmission capacity, leading to a lower network latency in the LM-embedded multiple-input multiple-output (MIMO) network. To design the FA-assisted LM-embedded network with the lower latency and higher accuracy requirements, the latency and peak signal to-noise ratio (PSNR) are considered in the objective function. Then, an efficient optimization algorithm is proposed under the block coordinate descent framework. Simulation results are provided to show the convergence behavior of the proposed algorithm, and the performance gains from the proposed FA-assisted LM-embedded network over the other benchmark networks in terms of network latency and PSNR.

The ability of robots to interpret human instructions and execute manipulation tasks necessitates the availability of task-relevant tabletop scenes for training. However, traditional methods for creating these scenes rely on time-consuming manual layout design or purely randomized layouts, which are limited in terms of plausibility or alignment with the tasks. In this paper, we formulate a novel task, namely task-oriented tabletop scene generation, which poses significant challenges due to the substantial gap between high-level task instructions and the tabletop scenes. To support research on such a challenging task, we introduce \textbf{MesaTask-10K}, a large-scale dataset comprising approximately 10,700 synthetic tabletop scenes with \emph{manually crafted layouts} that ensure realistic layouts and intricate inter-object relations. To bridge the gap between tasks and scenes, we propose a \textbf{Spatial Reasoning Chain} that decomposes the generation process into object inference, spatial interrelation reasoning, and scene graph construction for the final 3D layout. We present \textbf{MesaTask}, an LLM-based framework that utilizes this reasoning chain and is further enhanced with DPO algorithms to generate physically plausible tabletop scenes that align well with given task descriptions. Exhaustive experiments demonstrate the superior performance of MesaTask compared to baselines in generating task-conforming tabletop scenes with realistic layouts.

The advancement of Embodied AI heavily relies on large-scale, simulatable 3D scene datasets characterized by scene diversity and realistic layouts.However, existing datasets typically suffer from limitations in data scale or diversity, sanitized layouts lacking small items, and severe object collisions.To address these shortcomings, we introduce \textbf{InternScenes}, a novel large-scale simulatable indoor scene dataset comprising approximately 40,000 diverse scenes by integrating three disparate scene sources, \ie, real-world scans, procedurally generated scenes, and designer-created scenes, including 1.96M 3D objects and covering 15 common scene types and 288 object classes.We particularly preserve massive small items in the scenes, resulting in realistic and complex layouts with an average of 41.5 objects per region.Our comprehensive data processing pipeline ensures simulatability by creating real-to-sim replicas for real-world scans, enhances interactivity by incorporating interactive objects into these scenes, and resolves object collisions by physical simulations.We demonstrate the value of InternScenes with two benchmark applications: scene layout generation and point-goal navigation. Both show the new challenges posed by the complex and realistic layouts. More importantly, InternScenes paves the way for scaling up the model training for both tasks, making the generation and navigation in such complex scenes possible. We commit to open-sourcing the data and benchmarks to benefit the whole community.

We introduce AnySplat, a feed-forward network for novel-view synthesis from uncalibrated image collections. In contrast to traditional neural-rendering pipelines that demand known camera poses and per-scene optimization, or recent feed-forward methods that buckle under the computational weight of dense views—our model predicts everything in one shot. A single forward pass yields a set of 3D Gaussian primitives encoding both scene geometry and appearance, and the corresponding camera intrinsics and extrinsics for each input image. This unified design scales effortlessly to casually captured, multi-view datasets without any pose annotations. In extensive zero-shot evaluations, AnySplat matches the quality of pose-aware baselines in both sparse- and dense-view scenarios while surpassing existing pose-free approaches. Moreover, it greatly reduces rendering latency compared to optimization-based neural fields, bringing real-time novel-view synthesis within reach for unconstrained capture settings. Project page: https://city-super.github.io/anysplat/.