Searcharxiv⌕ Search

arXiv subjects

Siwei Zhang

Publications and source records attributed to Siwei Zhang.

At least 19 recordsLinked to original sources
arXiv2026

Electrostriction in a Bose-Einstein Condensate of Dipolar Molecules

The recent creation of a Bose-Einstein condensate (BEC) of dipolar molecules has opened a new frontier for many-body quantum systems in which dipolar interactions can drive novel self-organization phenomena. Here, we observe electrostriction in a molecular BEC, an elliptical deformation driven by anisotropic dipolar interactions. We use double microwave dressing, involving $\sigma$- and $\pi$-polarized fields, to control non-axially symmetric dipolar interactions. We compare the experimental observations of electrostriction to a model based on an extended Gross-Pitaevskii equation and find excellent agreement in the regime of weak to moderate interactions. Using electrostriction, we demonstrate that the molecular BEC can be torqued by dynamically changing the orientation of the elliptical $\sigma$ microwave field. This provides a route to setting molecular quantum gases into rotation, opening opportunities to probe vorticity, superfluidity, and supersolidity in strongly dipolar matter.

cond-mat.quant-gas↗
arXiv2026

Agent Lightning v1.0: Towards Harnessed Agentic RL

Modern agents operate inside agent harnesses that manage tools, context, and control flow, making the harness a critical part of the agent system. Our original Agent Lightning introduced a disaggregated architecture that connects arbitrary agents to RL training through an LLM endpoint proxy, an approach later adopted by frameworks such as verl Uni-Agent, AReaL 2.0, slime, and Polar. We refer to this paradigm as harnessed agentic RL, where the deploy-time harness directly participates in model post-training. Harnessed agentic RL differs fundamentally from traditional agentic RL: the harness, rather than the training engine, owns the environment interaction loop, while the trainer observes only sequences of LLM request-response pairs. This introduces challenges in retokenization, sample merging, advantage calculation, loss normalization, and backend scheduling, which can substantially affect training stability and effectiveness. We present Agent Lightning v1.0, a lightweight framework for harnessed agentic RL implemented in approximately 3,500 lines of code. It supports arbitrary agent harnesses and serves as a practical testbed for studying these challenges. We evaluate it on instruction-following, search, and coding agents, and provide a complete reproducible pipeline for coding-agent RL. Using only 6K training examples and modest compute, RL improves Qwen3.5-9B on SWE-bench Verified from 41.8% to 56.4%, a 14.6-point absolute gain. We release the complete workflow and training scripts to facilitate reproducible research on harnessed agentic RL.

cs.AI↗
arXiv2026

SVCBench: A Streaming Video Counting Benchmark for Spatial-Temporal State Maintenance

Video understanding requires models to continuously track and update world state during playback. Although existing benchmarks have advanced video understanding evaluation across multiple dimensions, they provide limited visibility into how models maintain world state over time. We propose SVCBench, a Streaming Video Counting Benchmark that repositions counting as a minimal, controlled probe for diagnosing models' world-state maintenance capability. We decompose this capability into object counting and event counting, forming 8 fine-grained subcategories. Object counting covers tracking currently visible objects and cumulative unique identities, while event counting covers detecting instantaneous actions and tracking complete activity cycles. SVCBench contains 406 videos with frame-by-frame annotations of 10,071 event occurrences and object state changes, yielding 1,000 streaming QA pairs with 4,576 query points distributed along video timelines. By observing state maintenance trajectories through streaming multi-point queries, we design three complementary metrics to diagnose numerical precision, trajectory consistency, and temporal awareness. Evaluations of mainstream video-language models show that current models still exhibit significant deficiencies in spatial-temporal state maintenance, with especially poor performance on periodic event counting. SVCBench provides a diagnostic framework for measuring and improving state maintenance in video understanding systems. Our code and data are available at https://buaa-colalab.github.io/SVCBench.

cs.CV↗
arXiv2026

RAPO: Expanding Exploration for LLM Agents via Retrieval-Augmented Policy Optimization

Agentic Reinforcement Learning (Agentic RL) has shown remarkable potential in large language model-based (LLM) agents. These works can empower LLM agents to tackle complex tasks via multi-step, tool-integrated reasoning. However, an inherent limitation of existing Agentic RL methods is their reliance on a pure on-policy paradigm for exploration, restricting exploration to the agent's self-generated outputs and preventing the discovery of new reasoning perspectives for further improvement. While recent efforts incorporate auxiliary off-policy signals to enhance exploration, they typically utilize full off-policy trajectories for trajectory-level policy estimation, overlooking the necessity for the fine-grained, step-level exploratory dynamics within agentic rollout. In this paper, we revisit exploration in Agentic RL and propose Retrieval-Augmented Policy Optimization (RAPO), a novel RL framework that introduces retrieval to explicitly expand exploration during training. To achieve this, we decompose the Agentic RL training process into two phases: (i) Hybrid-policy Agentic Rollout, and (ii) Retrieval-aware Policy Optimization. Specifically, we propose a Hybrid-policy Agentic Rollout strategy, which allows the agents to continuously reason over the retrieved off-policy step-level traces. It dynamically extends the reasoning receptive field of agents, enabling broader exploration conditioned on external behaviors. Subsequently, we introduce the Retrieval-aware Policy Optimization mechanism, which calibrates the policy gradient estimation with retrieval reward and importance shaping, stabilizing training and prioritizing retrieval-illuminating exploration. Extensive experiments show that RAPO achieves an +5.0% average gain on fourteen datasets across three agentic reasoning tasks, while delivering 1.2x faster training efficiency.

cs.AI↗
arXiv2026

SARAH: Spatially Aware Real-time Agentic Humans

As embodied agents become central to VR, telepresence, and digital human applications, their motion must go beyond speech-aligned gestures: agents should turn toward users, respond to their movement, and maintain natural gaze. Current methods lack this spatial awareness. We close this gap with the first real-time, fully causal method for spatially-aware conversational motion, deployable on a streaming VR headset. Given a user's position and dyadic audio, our approach produces full-body motion that aligns gestures with speech while orienting the agent according to the user. Our architecture combines a causal transformer-based VAE with interleaved latent tokens for streaming inference and a flow matching model conditioned on user trajectory and audio. To support varying gaze preferences, we introduce a gaze scoring mechanism with classifier-free guidance to decouple learning from control: the model captures natural spatial alignment from data, while users can adjust eye contact intensity at inference time. On the Embody 3D dataset, our method achieves state-of-the-art motion quality at over 300 FPS -- 3x faster than non-causal baselines -- while capturing the subtle spatial dynamics of natural conversation. We validate our approach on a live VR system, bringing spatially-aware conversational agents to real-time deployment. Please see https://evonneng.github.io/sarah/ for details.

cs.CV↗
arXiv2026

Masked Modeling for Human Motion Recovery Under Occlusions

Human motion reconstruction from monocular videos is a fundamental challenge in computer vision, with broad applications in AR/VR, robotics, and digital content creation, but remains challenging under frequent occlusions in real-world settings. Existing regression-based methods are efficient but fragile to missing observations, while optimization- and diffusion-based approaches improve robustness at the cost of slow inference speed and heavy preprocessing steps. To address these limitations, we leverage recent advances in generative masked modeling and present MoRo: Masked Modeling for human motion Recovery under Occlusions. MoRo is an occlusion-robust, end-to-end generative framework that formulates motion reconstruction as a video-conditioned task, and efficiently recover human motion in a consistent global coordinate system from RGB videos. By masked modeling, MoRo naturally handles occlusions while enabling efficient, end-to-end inference. To overcome the scarcity of paired video-motion data, we design a cross-modality learning scheme that learns multi-modal priors from a set of heterogeneous datasets: (i) a trajectory-aware motion prior trained on MoCap datasets, (ii) an image-conditioned pose prior trained on image-pose datasets, capturing diverse per-frame poses, and (iii) a video-conditioned masked transformer that fuses motion and pose priors, finetuned on video-motion datasets to integrate visual cues with motion dynamics for robust inference. Extensive experiments on EgoBody and RICH demonstrate that MoRo substantially outperforms state-of-the-art methods in accuracy and motion realism under occlusions, while performing on-par in non-occluded scenarios. MoRo achieves real-time inference at 70 FPS on a single H200 GPU.

cs.CV↗
arXiv2026

MMUEChange: A Generalized LLM Agent Framework for Intelligent Multi-Modal Urban Environment Change Analysis

Understanding urban environment change is essential for sustainable development. However, current approaches, particularly remote sensing change detection, often rely on rigid, single-modal analysis. To overcome these limitations, we propose MMUEChange, a multi-modal agent framework that flexibly integrates heterogeneous urban data via a modular toolkit and a core module, Modality Controller for cross- and intra-modal alignment, enabling robust analysis of complex urban change scenarios. Case studies include: a shift toward small, community-focused parks in New York, reflecting local green space efforts; the spread of concentrated water pollution across districts in Hong Kong, pointing to coordinated water management; and a notable decline in open dumpsites in Shenzhen, with contrasting links between nighttime economic activity and waste types, indicating differing urban pressures behind domestic and construction waste. Compared to the best-performing baseline, the MMUEChange agent achieves a 46.7% improvement in task success rate and effectively mitigates hallucination, demonstrating its capacity to support complex urban change analysis tasks with real-world policy implications.

cs.AI↗
arXiv2025

Embody 3D: A Large-scale Multimodal Motion and Behavior Dataset

The Codec Avatars Lab at Meta introduces Embody 3D, a multimodal dataset of 500 individual hours of 3D motion data from 439 participants collected in a multi-camera collection stage, amounting to over 54 million frames of tracked 3D motion. The dataset features a wide range of single-person motion data, including prompted motions, hand gestures, and locomotion; as well as multi-person behavioral and conversational data like discussions, conversations in different emotional states, collaborative activities, and co-living scenarios in an apartment-like space. We provide tracked human motion including hand tracking and body shape, text annotations, and a separate audio track for each participant.

cs.CV↗
arXiv2025

Observation of Self-Bound Droplets of Ultracold Dipolar Molecules

Ultracold gases of dipolar molecules have long been envisioned as a platform for the realization of novel quantum phases. Recent advances in collisional shielding, protecting molecules from inelastic losses, have enabled the creation of degenerate Fermi gases and, more recently, Bose-Einstein condensation of dipolar molecules. However, the observation of quantum phases in ultracold molecular gases that are driven by dipole-dipole interactions has so far remained elusive. In this work, we report the formation of self-bound droplets and droplet arrays in an ultracold gas of strongly dipolar sodium-cesium molecules. Starting from a molecular Bose-Einstein condensate (BEC), microwave dressing fields are used to induce dipole-dipole interactions with controllable strength and anisotropy. By varying the speed at which interactions are induced, covering a dynamic range of four orders of magnitude, we prepare droplets under equilibrium and non-equilibrium conditions, observing a transition from robust one-dimensional (1D) arrays to fluctuating two-dimensional (2D) structures. The droplets exhibit densities up to 100 times higher than the initial BEC, reaching the strongly interacting regime, and suggesting the possibility of a quantum-liquid or crystalline state. This work establishes ultracold molecules as a system for the exploration of strongly dipolar quantum matter and opens the door to the realization of self-organized crystal phases and dipolar spin liquids in optical lattices.

cond-mat.quant-gas↗
arXiv2025

VolumetricSMPL: A Neural Volumetric Body Model for Efficient Interactions, Contacts, and Collisions

Parametric human body models play a crucial role in computer graphics and vision, enabling applications ranging from human motion analysis to understanding human-environment interactions. Traditionally, these models use surface meshes, which pose challenges in efficiently handling interactions with other geometric entities, such as objects and scenes, typically represented as meshes or point clouds. To address this limitation, recent research has explored volumetric neural implicit body models. However, existing works are either insufficiently robust for complex human articulations or impose high computational and memory costs, limiting their widespread use. To this end, we introduce VolumetricSMPL, a neural volumetric body model that leverages Neural Blend Weights (NBW) to generate compact, yet efficient MLP decoders. Unlike prior approaches that rely on large MLPs, NBW dynamically blends a small set of learned weight matrices using predicted shape- and pose-dependent coefficients, significantly improving computational efficiency while preserving expressiveness. VolumetricSMPL outperforms prior volumetric occupancy model COAP with 10x faster inference, 6x lower GPU memory usage, enhanced accuracy, and a Signed Distance Function (SDF) for efficient and differentiable contact modeling. We demonstrate VolumetricSMPL's strengths across four challenging tasks: (1) reconstructing human-object interactions from in-the-wild images, (2) recovering human meshes in 3D scenes from egocentric views, (3) scene-constrained motion synthesis, and (4) resolving self-intersections. Our results highlight its broad applicability and significant performance and efficiency gains.

cs.CV↗
arXiv2025

Extreme Loss Suppression and Wide Tunability of Dipolar Interactions in an Ultracold Molecular Gas

Ultracold dipolar molecules hold great promise for the creation of novel quantum states of matter, but the realization of long-lived molecular bulk samples with strong dipole-dipole interactions has remained elusive. Here, we realize a collisionally stable gas of ultracold ground state molecules with a lifetime of several seconds. Utilizing double microwave dressing, we achieve an extreme suppression of inelastic two- and three-body losses by factors of more than 10,000 and 1,000, respectively. We find that losses remain suppressed across a wide range of dipole-dipole interactions, allowing the continuous tuning of the dipolar length from 0 to 1 um $\sim$ 20,000 $a_0$. Combined with the recent realization of Bose-Einstein condensation of dipolar molecules, our findings open the door to the exploration of strongly dipolar quantum liquids.

cond-mat.quant-gas↗
arXiv2025

Rethinking Time Encoding via Learnable Transformation Functions

Effectively modeling time information and incorporating it into applications or models involving chronologically occurring events is crucial. Real-world scenarios often involve diverse and complex time patterns, which pose significant challenges for time encoding methods. While previous methods focus on capturing time patterns, many rely on specific inductive biases, such as using trigonometric functions to model periodicity. This narrow focus on single-pattern modeling makes them less effective in handling the diversity and complexities of real-world time patterns. In this paper, we investigate to improve the existing commonly used time encoding methods and introduce Learnable Transformation-based Generalized Time Encoding (LeTE). We propose using deep function learning techniques to parameterize non-linear transformations in time encoding, making them learnable and capable of modeling generalized time patterns, including diverse and complex temporal dynamics. By enabling learnable transformations, LeTE encompasses previous methods as specific cases and allows seamless integration into a wide range of tasks. Through extensive experiments across diverse domains, we demonstrate the versatility and effectiveness of LeTE.

cs.LG↗
arXiv2025

Unifying Text Semantics and Graph Structures for Temporal Text-attributed Graphs with Large Language Models

Temporal graph neural networks (TGNNs) have shown remarkable performance in temporal graph modeling. However, real-world temporal graphs often possess rich textual information, giving rise to temporal text-attributed graphs (TTAGs). Such combination of dynamic text semantics and evolving graph structures introduces heightened complexity. Existing TGNNs embed texts statically and rely heavily on encoding mechanisms that biasedly prioritize structural information, overlooking the temporal evolution of text semantics and the essential interplay between semantics and structures for synergistic reinforcement. To tackle these issues, we present \textbf{CROSS}, a flexible framework that seamlessly extends existing TGNNs for TTAG modeling. CROSS is designed by decomposing the TTAG modeling process into two phases: (i) temporal semantics extraction; and (ii) semantic-structural information unification. The key idea is to advance the large language models (LLMs) to dynamically extract the temporal semantics in text space and then generate cohesive representations unifying both semantics and structures. Specifically, we propose a Temporal Semantics Extractor in the CROSS framework, which empowers LLMs to offer the temporal semantic understanding of node's evolving contexts of textual neighborhoods, facilitating semantic dynamics. Subsequently, we introduce the Semantic-structural Co-encoder, which collaborates with the above Extractor for synthesizing illuminating representations by jointly considering both semantic and structural information while encouraging their mutual reinforcement. Extensive experiments show that CROSS achieves state-of-the-art results on four public datasets and one industrial dataset, with 24.7% absolute MRR gain on average in temporal link prediction and 3.7% AUC gain in node classification of industrial application.

cs.CL↗
arXiv2025

Double Microwave Shielding

We develop double microwave shielding, which has recently enabled evaporative cooling to the first Bose-Einstein condensate of polar molecules [Bigagli et al., Nature 631, 289 (2024)]. Two microwave fields of different frequency and polarization are employed to effectively shield polar molecules from inelastic collisions and three-body recombination. Here, we describe in detail the theory of double microwave shielding. We demonstrate that double microwave shielding effectively suppresses two- and three-body losses. Simultaneously, dipolar interactions and the scattering length can be flexibly tuned, enabling comprehensive control over interactions in ultracold gases of polar molecules. We show that this approach works universally for a wide range of molecules. This opens the door to studying many-body physics with strongly interacting dipolar quantum matter.

cond-mat.quant-gas↗
arXiv2024

LLM-GAN: Construct Generative Adversarial Network Through Large Language Models For Explainable Fake News Detection

Explainable fake news detection predicts the authenticity of news items with annotated explanations. Today, Large Language Models (LLMs) are known for their powerful natural language understanding and explanation generation abilities. However, presenting LLMs for explainable fake news detection remains two main challenges. Firstly, fake news appears reasonable and could easily mislead LLMs, leaving them unable to understand the complex news-faking process. Secondly, utilizing LLMs for this task would generate both correct and incorrect explanations, which necessitates abundant labor in the loop. In this paper, we propose LLM-GAN, a novel framework that utilizes prompting mechanisms to enable an LLM to become Generator and Detector and for realistic fake news generation and detection. Our results demonstrate LLM-GAN's effectiveness in both prediction performance and explanation quality. We further showcase the integration of LLM-GAN to a cloud-native AI platform to provide better fake news detection service in the cloud.

cs.CL↗
arXiv2024

DTFormer: A Transformer-Based Method for Discrete-Time Dynamic Graph Representation Learning

Discrete-Time Dynamic Graphs (DTDGs), which are prevalent in real-world implementations and notable for their ease of data acquisition, have garnered considerable attention from both academic researchers and industry practitioners. The representation learning of DTDGs has been extensively applied to model the dynamics of temporally changing entities and their evolving connections. Currently, DTDG representation learning predominantly relies on GNN+RNN architectures, which manifest the inherent limitations of both Graph Neural Networks (GNNs) and Recurrent Neural Networks (RNNs). GNNs suffer from the over-smoothing issue as the models architecture goes deeper, while RNNs struggle to capture long-term dependencies effectively. GNN+RNN architectures also grapple with scaling to large graph sizes and long sequences. Additionally, these methods often compute node representations separately and focus solely on individual node characteristics, thereby overlooking the behavior intersections between the two nodes whose link is being predicted, such as instances where the two nodes appear together in the same context or share common neighbors. This paper introduces a novel representation learning method DTFormer for DTDGs, pivoting from the traditional GNN+RNN framework to a Transformer-based architecture. Our approach exploits the attention mechanism to concurrently process topological information within the graph at each timestamp and temporal dynamics of graphs along the timestamps, circumventing the aforementioned fundamental weakness of both GNNs and RNNs. Moreover, we enhance the model's expressive capability by incorporating the intersection relationships among nodes and integrating a multi-patching module. Extensive experiments conducted on six public dynamic graph benchmark datasets confirm our model's efficacy, achieving the SOTA performance.

cs.LG↗
arXiv2024

Three-Body Recombination of Ultracold Microwave-Shielded Polar Molecules

A combined experimental and theoretical study is carried out on the three-body recombination process in a gas of microwave-shielded polar molecules. For ground-state polar molecules dressed with a strong microwave field, field-linked bound states can appear in the intermolecular potential. We model three-body recombination into such bound states using classical trajectory calculations. Our results show that recombination can explain the enhanced loss rates observed at small microwave detunings in trapped samples of bosonic NaCs [Bigagli, $\textit{et al.}$, Nat. Phys. $\textbf{19}$ 1579-1584 (2023)]. Specifically, our calculations reproduce the experimentally measured three-body loss rates across a wide range of microwave Rabi couplings, detunings, and temperatures. This work suggests that for bosonic shielded molecular systems in which the two-body loss is sufficiently suppressed and a field-linked bound state is present, the dominant loss process will be three-body recombination.

cond-mat.quant-gas↗
arXiv2024

Dressed-State Spectroscopy and Magic Trapping of Microwave-Shielded NaCs Molecules

We report on the optical polarizability of microwave-shielded ultracold NaCs molecules in an optical dipole trap. While dressing a pair of rotational states with a microwave field, we observe a marked dependence of the optical polarizability on the intensity and detuning of the dressing field. To precisely characterize differential energy shifts between dressed rotational states, we establish dressed-state spectroscopy. For strong dressing fields, we find that a magic rotational transition can be engineered and demonstrate its insensitivity to laser intensity fluctuations. The results of this work have direct relevance for evaporative cooling and the recent demonstration of molecular Bose-Einstein condensates [Bigagli, et al., Nature (2024)] and may open a door to precision microwave spectroscopy in interacting many-body systems of microwave-shielded molecules.

cond-mat.quant-gas↗