arXiv subjects

Yue Pan

Publications and source records attributed to Yue Pan.

At least 55 records · Page 3Linked to original sources

arXiv2025

CHIME: Chiplet-based Heterogeneous Near-Memory Acceleration for Edge Multimodal LLM Inference

The proliferation of large language models (LLMs) is accelerating the integration of multimodal assistants into edge devices, where inference is executed under stringent latency and energy constraints, often exacerbated by intermittent connectivity. These challenges become particularly acute in the context of multimodal LLMs (MLLMs), as high-dimensional visual inputs are transformed into extensive token sequences, thereby inflating the key-value (KV) cache and imposing substantial data movement overheads to the LLM backbone. To address these issues, we present CHIME, a chiplet-based heterogeneous near-memory acceleration for edge MLLMs inference. CHIME leverages the complementary strengths of integrated monolithic 3D (M3D) DRAM and RRAM chiplets: DRAM supplies low-latency bandwidth for attention, while RRAM offers dense, non-volatile storage for weights. This heterogeneous hardware is orchestrated by a co-designed mapping framework that executes fused kernels near data, minimizing cross-chiplet traffic to maximize effective bandwidth. On FastVLM (0.6B/1.7B) and MobileVLM (1.7B/3B), CHIME achieves up to 54x speedup and up to 246x better energy efficiency per inference as compared to the edge GPU NVIDIA Jetson Orin NX. It sustains 116.5-266.5 token/J compared to Jetson's 0.7-1.1 token/J. Furthermore, it delivers up to 69.2x higher throughput than the state-of-the-art PIM accelerator FACIL. Compared to the M3D DRAM-only design, CHIME's heterogeneous memory further improves energy efficiency by 7% and performance by 2.4x.

cs.AR↗

arXiv2025

Stratum: System-Hardware Co-Design with Tiered Monolithic 3D-Stackable DRAM for Efficient MoE Serving

As Large Language Models (LLMs) continue to evolve, Mixture of Experts (MoE) architecture has emerged as a prevailing design for achieving state-of-the-art performance across a wide range of tasks. MoE models use sparse gating to activate only a handful of expert sub-networks per input, achieving billion-parameter capacity with inference costs akin to much smaller models. However, such models often pose challenges for hardware deployment due to the massive data volume introduced by the MoE layers. To address the challenges of serving MoE models, we propose Stratum, a system-hardware co-design approach that combines the novel memory technology Monolithic 3D-Stackable DRAM (Mono3D DRAM), near-memory processing (NMP), and GPU acceleration. The logic and Mono3D DRAM dies are connected through hybrid bonding, whereas the Mono3D DRAM stack and GPU are interconnected via silicon interposer. Mono3D DRAM offers higher internal bandwidth than HBM thanks to the dense vertical interconnect pitch enabled by its monolithic structure, which supports implementations of higher-performance near-memory processing. Furthermore, we tackle the latency differences introduced by aggressive vertical scaling of Mono3D DRAM along the z-dimension by constructing internal memory tiers and assigning data across layers based on access likelihood, guided by topic-based expert usage prediction to boost NMP throughput. The Stratum system achieves up to 8.29x improvement in decoding throughput and 7.66x better energy efficiency across various benchmarks compared to GPU baselines.

cs.AR↗

arXiv2025

Towards Speeding up Program Repair with Non-Autoregressive Model

Enlightened by the success of machine learning techniques in various application areas, recent years have witnessed a surge of research efforts on automatic program repair (APR) using machine learning techniques. Previous machine learning-based APR techniques essentially modified bugs in the autoregressive (AR) manner, which predicts future values based on past values. Due to the manner of token-by-token generation, the AR-based APR technique has a huge time delay. In particular, the delay of the APR model with a large number of parameters is more serious. To address the issue, we aim to apply the non-autoregressive (NAR) method to the APR task, which can output target code in a parallel manner to avoid huge repair delays. However, the naive use of the NAR manner for the APR task suffers from the issue of compromised patch quality. To effectively adapt the NAR manner for the APR task, we in this paper propose NARRepair, the first customized NAR code generation model for the APR task. The NARRepair model features three major novelties, including 1) the repair action predictor for alleviating the over-correction issue, 2) the inter-token dependency extractor for alleviating the issue of lacking inter-token dependency information, and 3) the two-stage decoder for alleviating the issue of lacking contextual information. We evaluated NARRepair on three widely used datasets in the APR community, and the results show that 1) compared to other APR techniques, the NARRepair model has the best performance within the limited repair time, and 2) compared to AR-based APR techniques, the repair speed of NARRepair has been increased by 1.4-6.4 times in the GPU environment. Overall, the results show that NARRepair has achieved state-of-the-art comprehensive performance in terms of repair speed and accuracy.

cs.SE↗

arXiv2025

FastCuRL: Curriculum Reinforcement Learning with Stage-wise Context Scaling for Efficient Training R1-like Reasoning Models

Improving training efficiency continues to be one of the primary challenges in large-scale Reinforcement Learning (RL). In this paper, we investigate how context length and the complexity of training data influence the RL scaling training process of R1-distilled reasoning models, e.g., DeepSeek-R1-Distill-Qwen-1.5B. Our experimental results reveal that: (1) simply controlling the context length and curating the training data based on the input prompt length can effectively improve the training efficiency of RL scaling, achieving better performance with more concise CoT; (2) properly scaling the context length helps mitigate entropy collapse; and (3) carefully choosing the context length facilitates achieving efficient LLM training and reasoning. Inspired by these insights, we propose FastCuRL, a curriculum RL framework with stage-wise context scaling to achieve efficient LLM training and reasoning. Extensive experimental results demonstrate that FastCuRL-1.5B-V3 significantly outperforms state-of-the-art reasoning models on five competition-level benchmarks and achieves 49.6% accuracy on AIME 2024. Furthermore, FastCuRL-1.5B-Preview surpasses DeepScaleR-1.5B-Preview on five benchmarks while only using a single node with 8 GPUs and a total of 50% of training steps.

cs.CL↗

arXiv2025

PINGS: Gaussian Splatting Meets Distance Fields within a Point-Based Implicit Neural Map

Robots benefit from high-fidelity reconstructions of their environment, which should be geometrically accurate and photorealistic to support downstream tasks. While this can be achieved by building distance fields from range sensors and radiance fields from cameras, realising scalable incremental mapping of both fields consistently and at the same time with high quality is challenging. In this paper, we propose a novel map representation that unifies a continuous signed distance field and a Gaussian splatting radiance field within an elastic and compact point-based implicit neural map. By enforcing geometric consistency between these fields, we achieve mutual improvements by exploiting both modalities. We present a novel LiDAR-visual SLAM system called PINGS using the proposed map representation and evaluate it on several challenging large-scale datasets. Experimental results demonstrate that PINGS can incrementally build globally consistent distance and radiance fields encoded with a compact set of neural points. Compared to state-of-the-art methods, PINGS achieves superior photometric and geometric rendering at novel views by constraining the radiance field with the distance field. Furthermore, by utilizing dense photometric cues and multi-view consistency from the radiance field, PINGS produces more accurate distance fields, leading to improved odometry estimation and mesh reconstruction. We also provide an open-source implementation of PING at: https://github.com/PRBonn/PINGS.

cs.RO↗

arXiv2025

Much Ado About Nothing: Galaxy Formation and Galactic Outflows in Cosmic Voids

We present a theoretical framework for calculating the volume filling fraction of galactic outflows in cosmic voids by integrating analytical models for the halo mass function (HMF), the halo occupation fraction, the stellar mass-halo mass relation, and outflow sizes. Using RAMSES, we perform a hydrodynamical zoom-in simulation of the central 25 cMpc/h region of a spherical void, identified as the lowest-density region among 1,000 random spheres in a parent 1 Gpc box simulation. This void has a diameter of 120 cMpc/h and a density contrast of $δ\simeq -0.8$. We find that the properties of void galaxies remain stable when expanding the zoom-in region to 50 cMpc/h, though our relatively low mass resolution impacts the results. Our higher-resolution simulation aligns with the analytical HMF that accounts for the void's underdensity and size. While higher resolution improves stellar mass estimates for low-mass halos, computational constraints necessitate a theoretical framework that enables extrapolation to infinite resolution. Our analytical model, calibrated to our simulations, enables extrapolation down to the filtering mass of star-forming halos. To compare galaxy properties in this void with those in the field, we conduct a companion field simulation of the same box size. At infinite resolution, we predict wind volume filling fractions of $18.6\%$ in the field and $3.1\%$ in our void, with values dependent on cosmic variance, void size, and underdensity. Dwarf galaxies contribute minimally, and resolving halos to $M_{\rm h}=10^{10} M_\odot$ suffices for robust estimates. Applying our framework to the Local Group void ($δ\simeq -0.5$, $R=20\ \mathrm{cMpc}$), we predict a wind volume filling fraction of $9.6\%\pm3.3\%$.

astro-ph.GA↗

arXiv2025

A Chinese Heart Failure Status Speech Database with Universal and Personalised Classification

Speech is a cost-effective and non-intrusive data source for identifying acute and chronic heart failure (HF). However, there is a lack of research on whether Chinese syllables contain HF-related information, as observed in other well-studied languages. This study presents the first Chinese speech database of HF patients, featuring paired recordings taken before and after hospitalisation. The findings confirm the effectiveness of the Chinese language in HF detection using both standard 'patient-wise' and personalised 'pair-wise' classification approaches, with the latter serving as an ideal speaker-decoupled baseline for future research. Statistical tests and classification results highlight individual differences as key contributors to inaccuracy. Additionally, an adaptive frequency filter (AFF) is proposed for frequency importance analysis. The data and demonstrations are published at https://github.com/panyue1998/Voice_HF.

eess.AS↗

arXiv2025

Amico: An Event-Driven Modular Framework for Persistent and Embedded Autonomy

Recent advances in large language models (LLMs) and autonomous agents have enabled systems capable of performing complex tasks across domains such as human-computer interaction, planning, and web navigation. However, many existing frameworks struggle in real-world or resource-constrained environments due to their reliance on cloud-based computation, limited robustness in dynamic contexts, and lack of persistent autonomy and environmental awareness. We present Amico, a modular, event-driven framework for building autonomous agents optimized for embedded systems. Written in Rust for safety and performance, Amico supports reactive, persistent agents that operate efficiently across embedded platforms and browser environments via WebAssembly. It provides clean abstractions for event handling, state management, behavior execution, and integration with reasoning modules. Amico delivers a unified infrastructure for constructing resilient, interactive agents suitable for deployment in settings with limited compute and intermittent connectivity.

cs.AI↗

arXiv2025

The Hitchhiker's Guide to Programming and Optimizing Cache Coherent Heterogeneous Systems: CXL, NVLink-C2C, and AMD Infinity Fabric

We present a thorough analysis of the use of modern heterogeneous systems interconnected by various cachecoherent links, including CXL, NVLink-C2C, and Infinity Fabric. We studied a wide range of server systems that combined CPUs from different vendors and various types of coherent memory devices, including CXL memory expander, CXL pool, CXL shared memory, GH200 GPU, and AMD MI300a HBM. For this study, we developed a heterogeneous memory benchmark suite, Heimdall, to profile the performance of such heterogeneous systems and present a detailed performance comparison across systems. By leveraging H E I M DA L L , we unveiled the detailed architecture design in these systems, drew observations on optimizing performance for workloads, and pointed out directions for future development of cache coherent heterogeneous systems.

cs.PF↗

arXiv2025

Looping metal-support interaction in heterogeneous catalysts during redox reactions

Metal-support interfaces fundamentally govern the catalytic performance of heterogeneous systems through complex interactions. Here, utilizing operando transmission electron microscopy, we uncovered a type of looping metal-support interaction in NiFe-Fe3O4 catalysts during hydrogen oxidation reaction. At the NiFe-Fe3O4 interfaces, lattice oxygens react with NiFe-activated H atoms, gradually sacrificing themselves and resulting in dynamically migrating interfaces. Meanwhile, reduced iron atoms migrate to the {111} surface of Fe3O4 support and react with oxygen molecules. Consequently, the hydrogen oxidation reaction separates spatially on a single nanoparticle and is intrinsically coupled with the redox reaction of the Fe3O4 support through the dynamic migration of metal-support interfaces. Our work provides previously unidentified mechanistic insight into metal-support interactions and underscores the transformative potential of operando methodologies for studying atomic-scale dynamics.

cond-mat.mtrl-sci↗

arXiv2025

Globally Consistent RGB-D SLAM with 2D Gaussian Splatting

Recently, 3D Gaussian splatting-based RGB-D SLAM displays remarkable performance of high-fidelity 3D reconstruction. However, the lack of depth rendering consistency and efficient loop closure limits the quality of its geometric reconstructions and its ability to perform globally consistent mapping online. In this paper, we present 2DGS-SLAM, an RGB-D SLAM system using 2D Gaussian splatting as the map representation. By leveraging the depth-consistent rendering property of the 2D variant, we propose an accurate camera pose optimization method and achieve geometrically accurate 3D reconstruction. In addition, we implement efficient loop detection and camera relocalization by leveraging MASt3R, a 3D foundation model, and achieve efficient map updates by maintaining a local active map. Experiments show that our 2DGS-SLAM approach achieves superior tracking accuracy, higher surface reconstruction quality, and more consistent global map reconstruction compared to existing rendering-based SLAM methods, while maintaining high-fidelity image rendering and improved computational efficiency.

cs.RO↗

arXiv2025

Differentiation of Distinct Single Atoms via Multi-Defocus Fusion Method

High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) is a vital tool for characterizing single-atom catalysts (SACs). However, reliable elemental identification of different atoms remains challenging because the signal intensity of HAADF depends strongly on defocus and other imaging parameters, potentially ruining the Z-contrast of atoms at different depths. In this work, we investigated the influence of the vertical position of atoms (defocus), support thickness, interatomic height, convergence and collection angles via multi-slice simulations on a model system of Fe/Pt atoms on amorphous carbon supports. Our calculation shows that at a convergence angle of 28 mrad, a defocus of 4.6 nm can cause Fe and Pt atoms indistinguishable. At a larger convergence angle, this critical indistinguishable defocus can be even shorter. To address this limitation, we propose a Multi-Defocus Fusion (MDF) method, retrieving the Z-contrast from serial images from multiple defocus. Experimental validation on a Fe/Pt SAC sample confirms the effectiveness of MDF, yielding clearly separated intensity histograms corresponding to Fe and Pt atoms. This work presents a robust, easy-to-implement strategy for accurate single-atom identification, offering valuable guidance for the accelerated screening and rational design of high-performance SACs.

cond-mat.mtrl-sci↗

arXiv2025

ActiveGS: Active Scene Reconstruction Using Gaussian Splatting

Robotics applications often rely on scene reconstructions to enable downstream tasks. In this work, we tackle the challenge of actively building an accurate map of an unknown scene using an RGB-D camera on a mobile platform. We propose a hybrid map representation that combines a Gaussian splatting map with a coarse voxel map, leveraging the strengths of both representations: the high-fidelity scene reconstruction capabilities of Gaussian splatting and the spatial modelling strengths of the voxel map. At the core of our framework is an effective confidence modelling technique for the Gaussian splatting map to identify under-reconstructed areas, while utilising spatial information from the voxel map to target unexplored areas and assist in collision-free path planning. By actively collecting scene information in under-reconstructed and unexplored areas for map updates, our approach achieves superior Gaussian splatting reconstruction results compared to state-of-the-art approaches. Additionally, we demonstrate the real-world applicability of our framework using an unmanned aerial vehicle.

cs.RO↗

arXiv2025

Improving Indoor Localization Accuracy by Using an Efficient Implicit Neural Map Representation

Globally localizing a mobile robot in a known map is often a foundation for enabling robots to navigate and operate autonomously. In indoor environments, traditional Monte Carlo localization based on occupancy grid maps is considered the gold standard, but its accuracy is limited by the representation capabilities of the occupancy grid map. In this paper, we address the problem of building an effective map representation that allows to accurately perform probabilistic global localization. To this end, we propose an implicit neural map representation that is able to capture positional and directional geometric features from 2D LiDAR scans to efficiently represent the environment and learn a neural network that is able to predict both, the non-projective signed distance and a direction-aware projective distance for an arbitrary point in the mapped environment. This combination of neural map representation with a light-weight neural network allows us to design an efficient observation model within a conventional Monte Carlo localization framework for pose estimation of a robot in real time. We evaluated our approach to indoor localization on a publicly available dataset for global localization and the experimental results indicate that our approach is able to more accurately localize a mobile robot than other localization approaches employing occupancy or existing neural map representations. In contrast to other approaches employing an implicit neural map representation for 2D LiDAR localization, our approach allows to perform real-time pose tracking after convergence and near real-time global localization. The code of our approach is available at: https://github.com/PRBonn/enm-mcl.

cs.RO↗

arXiv2025

A Dataset and Benchmark for Shape Completion of Fruits for Agricultural Robotics

As the world population is expected to reach 10 billion by 2050, our agricultural production system needs to double its productivity despite a decline of human workforce in the agricultural sector. Autonomous robotic systems are one promising pathway to increase productivity by taking over labor-intensive manual tasks like fruit picking. To be effective, such systems need to monitor and interact with plants and fruits precisely, which is challenging due to the cluttered nature of agricultural environments causing, for example, strong occlusions. Thus, being able to estimate the complete 3D shapes of objects in presence of occlusions is crucial for automating operations such as fruit harvesting. In this paper, we propose the first publicly available 3D shape completion dataset for agricultural vision systems. We provide an RGB-D dataset for estimating the 3D shape of fruits. Specifically, our dataset contains RGB-D frames of single sweet peppers in lab conditions but also in a commercial greenhouse. For each fruit, we additionally collected high-precision point clouds that we use as ground truth. For acquiring the ground truth shape, we developed a measuring process that allows us to record data of real sweet pepper plants, both in the lab and in the greenhouse with high precision, and determine the shape of the sensed fruits. We release our dataset, consisting of almost 7,000 RGB-D frames belonging to more than 100 different fruits. We provide segmented RGB-D frames, with camera intrinsics to easily obtain colored point clouds, together with the corresponding high-precision, occlusion-free point clouds obtained with a high-precision laser scanner. We additionally enable evaluation of shape completion approaches on a hidden test set through a public challenge on a benchmark server.

cs.CV↗

arXiv2025

Can Many-Shot In-Context Learning Help LLMs as Evaluators? A Preliminary Empirical Study

Utilizing Large Language Models (LLMs) as evaluators to assess the performance of LLMs has garnered attention. However, this kind of evaluation approach is affected by potential biases within LLMs, raising concerns about the accuracy and reliability of the evaluation results of LLMs. To address this problem, we propose and study two many-shot In-Context Learning (ICL) prompt templates to help LLM evaluators mitigate potential biases: Many-Shot with Reference (MSwR) and Many-Shot without Reference (MSoR). Specifically, the former utilizes in-context examples with model-generated evaluation rationales as references, while the latter does not include these references. Using these prompt designs, we investigate the impact of increasing the number of in-context examples on the consistency and quality of the evaluation results. Experimental results show that advanced LLMs, such as GPT-4o, perform better in the many-shot regime than in the zero-shot and few-shot regimes. Furthermore, when using GPT-4o as an evaluator in the many-shot regime, adopting MSwR as the prompt template performs better than MSoR.

cs.CL↗

arXiv2025

Evidence of P-wave Pairing in K$_2$Cr$_3$As$_3$ Superconductors from Phase-sensitive Measurement

P-wave superconductors hold immense promise for both fundamental physics and practical applications due to their unusual pairing symmetry and potential topological superconductivity. However, the exploration of the p-wave superconductors has proved to be a complex endeavor. Not only are they rare in nature but also the identification of p-wave superconductors has been an arduous task in history. For example, phase-sensitive measurement, an experimental technique which can provide conclusive evidence for unconventional pairing, has not been implemented successfully to identify p-wave superconductors. Here, we study a recently discovered family of superconductors, A$_2$Cr$_3$As$_3$ (A = K, Rb, Cs), which were proposed theoretically to be a candidate of p-wave superconductors. We fabricate superconducting quantum interference devices (SQUIDs) on exfoliated K$_2$Cr$_3$As$_3$, and perform the phase-sensitive measurement. We observe that such SQUIDs exhibit a pronounced second-order harmonic component sin(2$π$) in the current-phase relation, suggesting the admixture of 0- and $π$-phase. By carefully examining the magnetic field dependence of the oscillation patterns of critical current and Shapiro steps under microwave irradiation, we reveal a crossover from 0- to $π$-dominating phase state and conclude that the existence of the $π$-phase is in favor of the p-wave pairing symmetry in K$_2$Cr$_3$As$_3$.

cond-mat.supr-con↗

arXiv2025

COOL-LAMPS. VII. Quantifying Strong-lens Scaling Relations with 177 Cluster-scale Strong Gravitational Lenses in DECaLS

We estimate the Einstein-radius-enclosed total mass for 177 cluster-scale strong gravitational lenses identified by the ChicagO Optically selected Lenses Located At the Margins of Public Surveys (COOL-LAMPS) collaboration with lens redshifts ranging from $0.2 \lessapprox z \lessapprox 1.0$ using the brightest-cluster-galaxy (BCG) redshift and an observable proxy for the Einstein radius. We constrain the Einstein-radius-enclosed luminosity and stellar mass by fitting parametric spectral energy distributions to aperture photometry from the Dark Energy Camera Legacy Survey in the $g$-, $r$-, and $z$-band Dark Energy Camera filters. We find that the BCG redshift, enclosed total mass, and enclosed luminosity are strongly correlated and well described by a planar relationship in 3D space. We find that the enclosed total mass and stellar mass are correlated with a logarithmic slope of $0.500^{+0.029}_{-0.031}$, and the enclosed total mass and stellar-to-total mass fraction are correlated with a logarithmic slope of $-0.495^{+0.032}_{-0.033}$. In tandem with the small radii within which these slopes are constrained, this may suggest invariance in baryon conversion efficiency and feedback strength as a function of cluster-centric radii in galaxy clusters. Additionally, the correlations described here should have utility in ranking strong-lensing candidates in upcoming imaging surveys -- such as Rubin/Legacy Survey of Space and Time -- in which an algorithmic treatment of strong lenses will be needed due to the sheer volume of data these surveys will produce.

astro-ph.GA↗