arXiv subjects

Song Wang

Publications and source records attributed to Song Wang.

At least 181 records · Page 10Linked to original sources

arXiv2025

Are They All Good? Evaluating the Quality of CoTs in LLM-based Code Generation

Large language models (LLMs) have demonstrated impressive performance in code generation, particularly when augmented with chain-of-thought (CoT) prompting techniques. They break down requirements into intermediate reasoning steps, which act as design rationales to guide LLMs in writing code like human programmers. Thus, the quality of these steps is crucial for ensuring the correctness and reliability of the generated code. However, little is known about the quality of CoT generated by LLMs. To what extent can we trust the thoughts generated by LLMs? How good are they? This paper empirically explores the external and internal factors of why LLMs generate unsatisfactory CoTs by analyzing 1,023 failed code samples on two widely used code generation benchmarks. We also evaluate their impact on code generation performance by analyzing 210 CoT-code pairs and refining the unsatisfied CoTs by prompting LLMs. Our study reveals three key findings: (1) External factors (53.60%), such as unclear requirements and lack of context, mainly affect CoT quality, while internal factors (40.10%) stem from LLMs' misunderstanding prompts. (2) Even when CoTs are correct, 18.5% of the generated code contains errors due to instruction-following issues; conversely, 11.90% of correct code is paired with flawed CoTs. (3) Refining low-quality CoTs is feasible, i.e., LLMs improve when given detailed problem descriptions. These findings highlight key challenges in CoT-based code generation and suggest directions for improving LLM reasoning and reliability.

cs.SE↗

arXiv2025

Beyond One Shot, Beyond One Perspective: Cross-View and Long-Horizon Distillation for Better LiDAR Representations

LiDAR representation learning aims to extract rich structural and semantic information from large-scale, readily available datasets, reducing reliance on costly human annotations. However, existing LiDAR representation strategies often overlook the inherent spatiotemporal cues in LiDAR sequences, limiting their effectiveness. In this work, we propose LiMA, a novel long-term image-to-LiDAR Memory Aggregation framework that explicitly captures longer range temporal correlations to enhance LiDAR representation learning. LiMA comprises three key components: 1) a Cross-View Aggregation module that aligns and fuses overlapping regions across neighboring camera views, constructing a more unified and redundancy-free memory bank; 2) a Long-Term Feature Propagation mechanism that efficiently aligns and integrates multi-frame image features, reinforcing temporal coherence during LiDAR representation learning; and 3) a Cross-Sequence Memory Alignment strategy that enforces consistency across driving sequences, improving generalization to unseen environments. LiMA maintains high pretraining efficiency and incurs no additional computational overhead during downstream tasks. Extensive experiments on mainstream LiDAR-based perception benchmarks demonstrate that LiMA significantly improves both LiDAR semantic segmentation and 3D object detection. We hope this work inspires more effective pretraining paradigms for autonomous driving. The code has be made publicly accessible for future research.

cs.CV↗

arXiv2025

SAM4D: Segment Anything in Camera and LiDAR Streams

We present SAM4D, a multi-modal and temporal foundation model designed for promptable segmentation across camera and LiDAR streams. Unified Multi-modal Positional Encoding (UMPE) is introduced to align camera and LiDAR features in a shared 3D space, enabling seamless cross-modal prompting and interaction. Additionally, we propose Motion-aware Cross-modal Memory Attention (MCMA), which leverages ego-motion compensation to enhance temporal consistency and long-horizon feature retrieval, ensuring robust segmentation across dynamically changing autonomous driving scenes. To avoid annotation bottlenecks, we develop a multi-modal automated data engine that synergizes VFM-driven video masklets, spatiotemporal 4D reconstruction, and cross-modal masklet fusion. This framework generates camera-LiDAR aligned pseudo-labels at a speed orders of magnitude faster than human annotation while preserving VFM-derived semantic fidelity in point cloud representations. We conduct extensive experiments on the constructed Waymo-4DSeg, which demonstrate the powerful cross-modal segmentation ability and great potential in data annotation of proposed SAM4D.

cs.CV↗

arXiv2025

PAGENT: Learning to Patch Software Engineering Agents

LLM Agents produce patches automatically to resolve an issue. However, they can generate inaccurate patches. Little is known about the root causes behind those failed patches or how those could be fixed. This paper reports an empirical study of the failed patches generated by seven top LLM code agents. We collected 114 issues from the SWE-bench Lite dataset that remained unresolved across the agents. The seven agents produced a total of 769 failed patches for those issues, which we checked with a combination of GPT-4o and manual analysis. We present a taxonomy of the failure reasons across the patches. The taxonomy contains six categories, with several sub-categories under each category. For example, a frequently observed category is the inability of an LLM to correctly infer/produce the appropriate variable type in the produced patch. As a first step towards addressing such type-related errors, we designed PAGENT (Patch Agent). PAGENT utilizes program analysis techniques like CFG creation and exploration to infer the type of information of a patch. PAGENT does this by applying repository-level static code analysis techniques. Then, PAGENT refines the inferred type by further utilizing an LLM-based inference technique. We tested PAGENT on all 127 type-related failed patches from the top three agents in our study. PAGENT could fix 29 of the 127 failed patches.

cs.SE↗

arXiv2025

Inst3D-LMM: Instance-Aware 3D Scene Understanding with Multi-modal Instruction Tuning

Despite encouraging progress in 3D scene understanding, it remains challenging to develop an effective Large Multi-modal Model (LMM) that is capable of understanding and reasoning in complex 3D environments. Most previous methods typically encode 3D point and 2D image features separately, neglecting interactions between 2D semantics and 3D object properties, as well as the spatial relationships within the 3D environment. This limitation not only hinders comprehensive representations of 3D scene, but also compromises training and inference efficiency. To address these challenges, we propose a unified Instance-aware 3D Large Multi-modal Model (Inst3D-LMM) to deal with multiple 3D scene understanding tasks simultaneously. To obtain the fine-grained instance-level visual tokens, we first introduce a novel Multi-view Cross-Modal Fusion (MCMF) module to inject the multi-view 2D semantics into their corresponding 3D geometric features. For scene-level relation-aware tokens, we further present a 3D Instance Spatial Relation (3D-ISR) module to capture the intricate pairwise spatial relationships among objects. Additionally, we perform end-to-end multi-task instruction tuning simultaneously without the subsequent task-specific fine-tuning. Extensive experiments demonstrate that our approach outperforms the state-of-the-art methods across 3D scene understanding, reasoning and grounding tasks. Source code is available at https://github.com/hanxunyu/Inst3D-LMM

cs.CV↗

arXiv2025

Four ages of rotating stars in the rotation--activity relationship and gyrochronology

Gyrochronology and the rotation--activity relationship are standard techniques used to determine the evolution phase.The mismatch in the definition of the evolutionary phases has so far raised many issues in physics and mathematics and hampered the understanding of how the internal dynamo processes affect the observable properties.To address this problem, we seek a unified scheme that shows a one-to-one mapping from gyrochronology to the rotation--activity relationship.We obtained the chromospheric activity $R'_{\rm HK}$ of 6846 stars and their rotation periods to investigate the rotation--activity relationship.We apply a three-interval model to fit the relationship and find that it is best fitted by three intervals in the range of Ro$<0.7$.We associate those intervals to the convective, gap and interface phases of gyrochronology.Furthermore,we suggest an additional epoch at late times of the I phase.We further use the three-interval models to fit the period--activity relationship in temperature bins and determine the duration of the transition phase as a function of effective temperature.By comparing the critical temperature and period of the g-to-I transition with the slowly rotating sequence of 10 young open clusters whose ages range from 1 Myr to 2.5 Gyr, we conclude that our new model finds the pure I sequence without fast rotating outliers, which defines the zero-age I sequence (ZAIS).We propose that there is an ambiguous consensus on when the I sequence starts to work.This ambiguity is from the visually convergent sequence of the color--period diagrams in open clusters.This visually convergent sequence is younger than the ZAIS and is actually the pre-I sequence that can be associated with the stall of the spin-down. Our results unify the rotation--activity relationship and gyrochonology for the stellar evolution of low-mass stars, for which we coined the ``CgIW" scenario.

astro-ph.SR↗

arXiv2025

CXR-LT 2024: A MICCAI challenge on long-tailed, multi-label, and zero-shot disease classification from chest X-ray

The CXR-LT series is a community-driven initiative designed to enhance lung disease classification using chest X-rays (CXR). It tackles challenges in open long-tailed lung disease classification and enhances the measurability of state-of-the-art techniques. The first event, CXR-LT 2023, aimed to achieve these goals by providing high-quality benchmark CXR data for model development and conducting comprehensive evaluations to identify ongoing issues impacting lung disease classification performance. Building on the success of CXR-LT 2023, the CXR-LT 2024 expands the dataset to 377,110 chest X-rays (CXRs) and 45 disease labels, including 19 new rare disease findings. It also introduces a new focus on zero-shot learning to address limitations identified in the previous event. Specifically, CXR-LT 2024 features three tasks: (i) long-tailed classification on a large, noisy test set, (ii) long-tailed classification on a manually annotated "gold standard" subset, and (iii) zero-shot generalization to five previously unseen disease findings. This paper provides an overview of CXR-LT 2024, detailing the data curation process and consolidating state-of-the-art solutions, including the use of multimodal models for rare disease detection, advanced generative approaches to handle noisy labels, and zero-shot learning strategies for unseen diseases. Additionally, the expanded dataset enhances disease coverage to better represent real-world clinical settings, offering a valuable resource for future research. By synthesizing the insights and innovations of participating teams, we aim to advance the development of clinically realistic and generalizable diagnostic models for chest radiography.

cs.CV↗

arXiv2025

A Be star-black hole binary with a wide orbit from LAMOST time-domain survey

Binary systems consisting of an early type star and a black hole (BH) are crucial for understanding various astrophysical phenomena, particularly the origins of detected gravitational wave sources. Be binary systems are expected to represent a key evolutionary stage in hosting BHs. However, while hundreds of Be X-ray binaries are known, the only confirmed BH candidate in a Be binary remains highly controversial. We report the discovery of ALS 8814, a Be star-BH binary with a moderately eccentric ($e = 0.23$) and wide orbit ($P = 176.6$ days), revealed by the radial velocity (RV) measurement of the visible Be star. Our analysis, combining flux-calibrated spectra in the Balmer discontinuity region and spectral template matching, yields a mass of $11.2^{+1.4}_{-1.2}$ $M_\odot$ for the Be star. The minimum mass of the unseen companion, assuming an edge-on inclination ($i = 90^{\circ}$), is $9.8\pm 0.7\,M_\odot$. We rule out the presence of non-degenerate companions in ALS 8814, indicating that it can only be a BH. This discovery represents a robust case of a Be-BH binary, identified purely through precise RV measurements from a single set of lines. The extremely low peculiar velocity of ALS 8814 suggests that the BH is formed via a direct core-collapse with a negligible natal kick, implying an almost perfect alignment between the Be star's spin and the orbital plane. In this context, the binary's inclination angle is estimated to be 22$^{\circ}$-49$^{\circ}$ by analyzing the shallow double-peaked profile of the H$α$ emission line. This inclination range corresponds to a BH mass estimate between $15\,M_\odot$ and $58\,M_\odot$. As the only unambiguous Be-BH binary system known to date, ALS 8814 provides valuable constraints on the BH formation in a binary system with a high-mass companion.

astro-ph.SR↗

arXiv2025

PixelThink: Towards Efficient Chain-of-Pixel Reasoning

Existing reasoning segmentation approaches typically fine-tune multimodal large language models (MLLMs) using image-text pairs and corresponding mask labels. However, they exhibit limited generalization to out-of-distribution scenarios without an explicit reasoning process. Although recent efforts leverage reinforcement learning through group-relative policy optimization (GRPO) to enhance reasoning ability, they often suffer from overthinking - producing uniformly verbose reasoning chains irrespective of task complexity. This results in elevated computational costs and limited control over reasoning quality. To address this problem, we propose PixelThink, a simple yet effective scheme that integrates externally estimated task difficulty and internally measured model uncertainty to regulate reasoning generation within a reinforcement learning paradigm. The model learns to compress reasoning length in accordance with scene complexity and predictive confidence. To support comprehensive evaluation, we introduce ReasonSeg-Diff, an extended benchmark with annotated reasoning references and difficulty scores, along with a suite of metrics designed to assess segmentation accuracy, reasoning quality, and efficiency jointly. Experimental results demonstrate that the proposed approach improves both reasoning efficiency and overall segmentation performance. Our work contributes novel perspectives towards efficient and interpretable multimodal understanding. The code and model will be publicly available.

cs.CV↗

arXiv2025

Enhancing LLM-Based Code Generation with Complexity Metrics: A Feedback-Driven Approach

Automatic code generation has gained significant momentum with the advent of Large Language Models (LLMs) such as GPT-4. Although many studies focus on improving the effectiveness of LLMs for code generation, very limited work tries to understand the generated code's characteristics and leverage that to improve failed cases. In this paper, as the most straightforward characteristic of code, we investigate the relationship between code complexity and the success of LLM generated code. Using a large set of standard complexity metrics, we first conduct an empirical analysis to explore their correlation with LLM's performance on code generation (i.e., Pass@1). Using logistic regression models, we identify which complexity metrics are most predictive of code correctness. Building on these findings, we propose an iterative feedback method, where LLMs are prompted to generate correct code based on complexity metrics from previous failed outputs. We validate our approach across multiple benchmarks (i.e., HumanEval, MBPP, LeetCode, and BigCodeBench) and various LLMs (i.e., GPT-4o, GPT-3.5 Turbo, Llama 3.1, and GPT-o3 mini), comparing the results with two baseline methods: (a) zero-shot generation, and (b) iterative execution-based feedback without our code complexity insights. Experiment results show that our approach makes notable improvements, particularly with a smaller LLM (GPT3.5 Turbo), where, e.g., Pass@1 increased by 35.71% compared to the baseline's improvement of 12.5% on the HumanEval dataset. The study expands experiments to BigCodeBench and integrates the method with the Reflexion code generation agent, leading to Pass@1 improvements of 20% (GPT-4o) and 23.07% (GPT-o3 mini). The results highlight that complexity-aware feedback enhances both direct LLM prompting and agent-based workflows.

cs.SE↗

arXiv2025

Revisiting the Li abundances of Stars with and without Detected Planets from the High Resolution Spectroscopy

Whether the presence of planets affects the lithium (Li) abundance of their host stars is still an open question. To investigate the difference of the Li abundance between planet-host stars (HS) and isolated stars (IS) with no detected planets, we analyze a large sample of stars with temperatures ranging from 4600 to 6600 K and metallicity ranging from -0.55 to +0.50. The sample consists of 279 HS whose spectra were taken from the California-Kepler Survey (CKS), which followed up planets detected by Kepler, and 171 IS whose spectra were taken from the Keck archive. The non-local thermodynamic equilibrium (non-LTE) effects were taken into consideration. It is found that the distribution of Li abundances in both the HS and IS groups are generally consistent with each other. This suggests that the presence of Kepler-like planets does not have a significant impact on Li depletion. We also found that the non-LTE corrections can not be neglected for stars with A(Li) over ~ 2.5 dex.

astro-ph.SR↗

arXiv2025

PointLoRA: Low-Rank Adaptation with Token Selection for Point Cloud Learning

Self-supervised representation learning for point cloud has demonstrated effectiveness in improving pre-trained model performance across diverse tasks. However, as pre-trained models grow in complexity, fully fine-tuning them for downstream applications demands substantial computational and storage resources. Parameter-efficient fine-tuning (PEFT) methods offer a promising solution to mitigate these resource requirements, yet most current approaches rely on complex adapter and prompt mechanisms that increase tunable parameters. In this paper, we propose PointLoRA, a simple yet effective method that combines low-rank adaptation (LoRA) with multi-scale token selection to efficiently fine-tune point cloud models. Our approach embeds LoRA layers within the most parameter-intensive components of point cloud transformers, reducing the need for tunable parameters while enhancing global feature capture. Additionally, multi-scale token selection extracts critical local information to serve as prompts for downstream fine-tuning, effectively complementing the global context captured by LoRA. The experimental results across various pre-trained models and three challenging public datasets demonstrate that our approach achieves competitive performance with only 3.43% of the trainable parameters, making it highly effective for resource-constrained applications. Source code is available at: https://github.com/songw-zju/PointLoRA.

cs.CV↗

arXiv2025

MAPLE: Many-Shot Adaptive Pseudo-Labeling for In-Context Learning

In-Context Learning (ICL) empowers Large Language Models (LLMs) to tackle diverse tasks by incorporating multiple input-output examples, known as demonstrations, into the input of LLMs. More recently, advancements in the expanded context windows of LLMs have led to many-shot ICL, which uses hundreds of demonstrations and outperforms few-shot ICL, which relies on fewer examples. However, this approach is often hindered by the high cost of obtaining large amounts of labeled data. To address this challenge, we propose Many-Shot Adaptive Pseudo-LabEling, namely MAPLE, a novel influence-based many-shot ICL framework that utilizes pseudo-labeled samples to compensate for the lack of label information. We first identify a subset of impactful unlabeled samples and perform pseudo-labeling on them by querying LLMs. These pseudo-labeled samples are then adaptively selected and tailored to each test query as input to improve the performance of many-shot ICL, without significant labeling costs. Extensive experiments on real-world datasets demonstrate the effectiveness of our framework, showcasing its ability to enhance LLM adaptability and performance with limited labeled data.

cs.AI↗

arXiv2025

SecVulEval: Benchmarking LLMs for Real-World C/C++ Vulnerability Detection

Large Language Models (LLMs) have shown promise in software engineering tasks, but evaluating their effectiveness in vulnerability detection is challenging due to the lack of high-quality datasets. Most existing datasets are limited to function-level labels, ignoring finer-grained vulnerability patterns and crucial contextual information. Also, poor data quality such as mislabeling, inconsistent annotations, and duplicates can lead to inflated performance and weak generalization. Moreover, by including only the functions, these datasets miss broader program context, like data/control dependencies and interprocedural interactions, that are essential for accurately understanding real-world security flaws. Without this context, detection models are evaluated under unrealistic assumptions. To address these limitations, this paper introduces SecVulEval, a benchmark designed to support fine-grained evaluation of LLMs and other detection methods with rich contextual information. SecVulEval focuses on real-world C/C++ vulnerabilities at the statement level. This granularity enables more precise evaluation of a model's ability to localize vulnerabilities, beyond simple binary classification at the function level. By incorporating rich contextual information, SecVulEval sets a new standard for vulnerability detection benchmarks in realistic scenarios. This benchmark includes 25,440 function samples covering 5,867 unique CVEs in C/C++ projects from 1999 to 2024. We evaluated the SOTA LLMs with a multi-agent-based approach. The evaluation on our dataset shows that the models are still far from accurately predicting vulnerable statements in a given function. The best-performing Claude-3.7-Sonnet model achieves 23.83% F1-score for detecting vulnerable statements with correct reasoning. Finally, we analyze the LLM outputs and provide insights into their behavior in vulnerability detection for C/C++.

cs.SE↗

arXiv2025

Acoustic Characterization of the Resonator in the Chinese Transverse Flute (dizi)

The dizi is a traditional Chinese transverse flute and is most distinguished from the western flute by the presence of a hole covered by a wrinkled membrane. In this study, we analyze the linear acoustical behavior of the dizi resonator through a detailed acoustical model that incorporates drilled toneholes, back end-holes, membrane hole, and upstream embouchure hole. The input admittance of the dizi is measured and modeled using the Transfer Matrix Method (TMM) and Transfer Matrix Method with external Interactions (TMMI). In comparison to measurements, the TMMI is shown to more accurately model the dizi than the TMM when compared to measurements. Our analysis reveals that attaching the membrane shifts admittance peaks to lower frequencies, reduces their magnitude, and influences tuning and harmonicity for different peaks and fingerings. The study further shows that the upstream branch, which includes the embouchure hole, complicates the evaluation of the tonehole lattice cutoff frequency, suggesting that it may not need to be considered for flute instruments. Cutoff frequencies exhibit distinct groupings across fingerings, influenced by the different tonehole lattices in the dizi: finger-hole lattice and end-hole lattice.

physics.app-ph↗

arXiv2025

Offboard Occupancy Refinement with Hybrid Propagation for Autonomous Driving

Vision-based occupancy prediction, also known as 3D Semantic Scene Completion (SSC), presents a significant challenge in computer vision. Previous methods, confined to onboard processing, struggle with simultaneous geometric and semantic estimation, continuity across varying viewpoints, and single-view occlusion. Our paper introduces OccFiner, a novel offboard framework designed to enhance the accuracy of vision-based occupancy predictions. OccFiner operates in two hybrid phases: 1) a multi-to-multi local propagation network that implicitly aligns and processes multiple local frames for correcting onboard model errors and consistently enhancing occupancy accuracy across all distances. 2) the region-centric global propagation, focuses on refining labels using explicit multi-view geometry and integrating sensor bias, particularly for increasing the accuracy of distant occupied voxels. Extensive experiments demonstrate that OccFiner improves both geometric and semantic accuracy across various types of coarse occupancy, setting a new state-of-the-art performance on the SemanticKITTI dataset. Notably, OccFiner significantly boosts the performance of vision-based SSC models, achieving accuracy levels competitive with established LiDAR-based onboard SSC methods. Furthermore, OccFiner is the first to achieve automatic annotation of SSC in a purely vision-based approach. Quantitative experiments prove that OccFiner successfully facilitates occupancy data loop-closure in autonomous driving. Additionally, we quantitatively and qualitatively validate the superiority of the offboard approach on city-level SSC static maps. The source code will be made publicly available at https://github.com/MasterHow/OccFiner.

cs.CV↗

arXiv2025

MPRM: A Markov Path-based Rule Miner for Efficient and Interpretable Knowledge Graph Reasoning

Rule mining in knowledge graphs enables interpretable link prediction. However, deep learning-based rule mining methods face significant memory and time challenges for large-scale knowledge graphs, whereas traditional approaches, limited by rigid confidence metrics, incur high computational costs despite sampling techniques. To address these challenges, we propose MPRM, a novel rule mining method that models rule-based inference as a Markov chain and uses an efficient confidence metric derived from aggregated path probabilities, significantly lowering computational demands. Experiments on multiple datasets show that MPRM efficiently mines knowledge graphs with over a million facts, sampling less than 1% of facts on a single CPU in 22 seconds, while preserving interpretability and boosting inference accuracy by up to 11% over baselines.

cs.AI↗

arXiv2025

DPSeg: Dual-Prompt Cost Volume Learning for Open-Vocabulary Semantic Segmentation

Open-vocabulary semantic segmentation aims to segment images into distinct semantic regions for both seen and unseen categories at the pixel level. Current methods utilize text embeddings from pre-trained vision-language models like CLIP but struggle with the inherent domain gap between image and text embeddings, even after extensive alignment during training. Additionally, relying solely on deep text-aligned features limits shallow-level feature guidance, which is crucial for detecting small objects and fine details, ultimately reducing segmentation accuracy. To address these limitations, we propose a dual prompting framework, DPSeg, for this task. Our approach combines dual-prompt cost volume generation, a cost volume-guided decoder, and a semantic-guided prompt refinement strategy that leverages our dual prompting scheme to mitigate alignment issues in visual prompt generation. By incorporating visual embeddings from a visual prompt encoder, our approach reduces the domain gap between text and image embeddings while providing multi-level guidance through shallow features. Extensive experiments demonstrate that our method significantly outperforms existing state-of-the-art approaches on multiple public datasets.

cs.CV↗