arXiv subjects

Ge Li

Publications and source records attributed to Ge Li.

At least 73 records · Page 4Linked to original sources

arXiv2025

AdapTrack: Constrained Decoding without Distorting LLM's Output Intent

Language model-based code generation and completion tools have been widely adopted, but they may sometimes produce code that does not meet necessary constraints, such as syntactic correctness or API existence. Constrained decoding techniques are developed to help the model generate code adhering to the constraints by greedily eliminating generation options that violate constraints at each step of the generation process. However, there is a severe limitation of constrained decoding, that it distorts the model's output intent, forcing it to produce code that may satisfy the constraint but does not match the development intent and is therefore incorrect. In response to this challenge, we propose AdapTrack. By incorporating backtracking into the generation process, AdapTrack avoids distorting the output intent of the model, thereby producing results that are not only constraint-compliant but also more semantically aligned with model's output intent. On our synthetic API completion dataset, AdapTrack can achieve up to 360.87% improvement compared to constrained decoding; on the real-world API completion dataset we collect that exhibits similar issues, AdapTrack can achieve up to 38.93% improvement over constrained decoding; in general code genration benchmarks, compared to constrained decoding, AdapTrack can achieve up to 7.84% improvement on HumanEval, and up to 6.42% improvement on MBPP. This indicates that, simply by better adhering to the model's output intent, AdapTrack can achieve significant improvements. We provide a theoretical proof that the distribution produced by AdapTrack aligns with the model's distribution given the generated tokens, thereby ensuring that the model's output intent is not distorted. Experiments on DSL problems show that, compared to existing methods, our approach can provide generation results that are more consistent with the language model's distribution.

cs.SE↗

arXiv2025

MoRe-ERL: Learning Motion Residuals using Episodic Reinforcement Learning

We propose MoRe-ERL, a framework that combines Episodic Reinforcement Learning (ERL) and residual learning, which refines preplanned reference trajectories into safe, feasible, and efficient task-specific trajectories. This framework is general enough to incorporate into arbitrary ERL methods and motion generators seamlessly. MoRe-ERL identifies trajectory segments requiring modification while preserving critical task-related maneuvers. Then it generates smooth residual adjustments using B-Spline-based movement primitives to ensure adaptability to dynamic task contexts and smoothness in trajectory refinement. Experimental results demonstrate that residual learning significantly outperforms training from scratch using ERL methods, achieving superior sample efficiency and task performance. Hardware evaluations further validate the framework, showing that policies trained in simulation can be directly deployed in real-world systems, exhibiting a minimal sim-to-real gap.

cs.RO↗

arXiv2025

Flow4Agent: Long-form Video Understanding via Motion Prior from Optical Flow

Long-form video understanding has always been a challenging problem due to the significant redundancy in both temporal and spatial contents. This challenge is further exacerbated by the limited context length of Multimodal Large Language Models (MLLMs). To address this issue, many previous works have attempted to extract key video information, where the "key" is typically semantic-aware and heavily dependent on the CLIP model as prior. In this paper, we propose Flow4Agent, a novel framework that pioneeringly incorporates motion priors from optical flow to facilitate LLM-based long video understanding. Flow4Agent mitigates the redundancy in long videos at both temporal and spatial levels through two core modules: Temporal Granularity Optimization (TGO) adaptively refines framelevel hierarchies, which first leverages coarse flow priors to group similar visual contents and then applies semantic priors to filter out highly irrelevant scene information. Motion Token Pruning (MTP) further refines the intra-frame visual representations, pruning high-redundancy video tokens using fine-grained optical flow information. Extensive experiments demonstrate that our Flow4Agent outperforms existing methods across a wide range of video MLLM benchmarks, especially for hour-level video understanding tasks, achieving 64.7% on Video-MME, 71.4% on MLVU and 60.4% on LongVideoBench.

cs.CV↗

arXiv2025

GramTrans: A Better Code Representation Approach in Code Generation

Code generation has shown great promise in assisting software development. A fundamental yet underexplored question is how the choice of code representation affects model performance. While existing studies employ various representations, such as treating code as plain text, grammar rule sequences, or syntax tree sequences, they lack a principled understanding of the relationship between parsing difficulty and model effectiveness. This paper proposes a conjecture: the easier a representation is to parse, the better performance the model achieves. We formalize this idea using grammar classes, where representations in simpler classes (e.g., LL(1)) are easier to parse. Through a controlled experiment on a Python-based DSL, we show that parsing difficulty strongly correlates with model performance. Motivated by this finding, we present GramTrans, a general approach that automatically transforms a context-free language into a representation within the LL(1) class. GramTrans introduces a novel hierarchical conflict elimination algorithm, enabling a flexible trade-off between syntactic simplicity and token efficiency. We evaluate GramTrans on both Python and Java using three code generation models: StarCoder 1B, DeepSeek-Coder 1.3B, and Qwen2.5 1.5B. Across multiple benchmarks, GramTrans consistently delivers significant improvements over baseline representations. Furthermore, our analysis of existing representations reconfirms the strong alignment between parsing difficulty and model performance, providing additional support for the conjecture.

cs.SE↗

arXiv2025

A Survey on Code Generation with LLM-based Agents

Code generation agents powered by large language models (LLMs) are revolutionizing the software development paradigm. Distinct from previous code generation techniques, code generation agents are characterized by three core features. 1) Autonomy: the ability to independently manage the entire workflow, from task decomposition to coding and debugging. 2) Expanded task scope: capabilities that extend beyond generating code snippets to encompass the full software development lifecycle (SDLC). 3) Enhancement of engineering practicality: a shift in research emphasis from algorithmic innovation toward practical engineering challenges, such as system reliability, process management, and tool integration. This domain has recently witnessed rapid development and an explosion in research, demonstrating significant application potential. This paper presents a systematic survey of the field of LLM-based code generation agents. We trace the technology's developmental trajectory from its inception and systematically categorize its core techniques, including both single-agent and multi-agent architectures. Furthermore, this survey details the applications of LLM-based agents across the full SDLC, summarizes mainstream evaluation benchmarks and metrics, and catalogs representative tools. Finally, by analyzing the primary challenges, we identify and propose several foundational, long-term research directions for the future work of the field.

cs.SE↗

arXiv2025

SemGuard: Real-Time Semantic Evaluator for Correcting LLM-Generated Code

Large Language Models (LLMs) can translate natural language requirements into code, yet empirical analyses of representative models reveal that semantic errors-programs that compile but behave incorrectly-constitute the majority of observed faults (e.g., >60% on DeepSeek-Coder-6.7B and QwenCoder-7B). Post-hoc repair pipelines detect such faults only after execution, incurring latency, relying on incomplete test suites, and often mis-localizing the defect. Since semantic drift originates in the autoregressive decoding process, intervening while the code is being generated is a direct way to stop error propagation. Constrained-decoding approaches such as ROCODE attempt this, but still wait until the entire program runs to obtain feedback and use entropy heuristics that do not truly capture semantics. A more effective solution must inject semantic signals-early and precisely-into the decoding process.We present SemGuard, a semantic-evaluator-driven framework that performs real-time, line-level semantic supervision. To train the evaluator, we build SemDiff, the first dataset with fine-grained annotations that mark the exact line where a correct and an incorrect implementation diverge. The evaluator, once embedded in the LLM's decoder, flags deviations on partial code, rolls back to the faulty line, and guides regeneration-without executing the program or requiring test cases. Across four benchmarks, SemGuard consistently outperforms state-of-the-art baselines. It lowers the semantic error rate by 19.86% on SemDiff relative to ROCODE, and lifts Pass@1 by 48.92% on the real-world LiveCodeBench with CodeLlama-7B. Similar gains hold for StarCoder2-7B on MBPP and for DeepSeekCoder-6.7B on the Java benchmark SemDiff-Java, demonstrating model- and language-agnostic effectiveness.

cs.SE↗

arXiv2025

LongCat-Flash Technical Report

We introduce LongCat-Flash, a 560-billion-parameter Mixture-of-Experts (MoE) language model designed for both computational efficiency and advanced agentic capabilities. Stemming from the need for scalable efficiency, LongCat-Flash adopts two novel designs: (a) Zero-computation Experts, which enables dynamic computational budget allocation and activates 18.6B-31.3B (27B on average) per token depending on contextual demands, optimizing resource usage. (b) Shortcut-connected MoE, which enlarges the computation-communication overlap window, demonstrating notable gains in inference efficiency and throughput compared to models of a comparable scale. We develop a comprehensive scaling framework for large models that combines hyperparameter transfer, model-growth initialization, a multi-pronged stability suite, and deterministic computation to achieve stable and reproducible training. Notably, leveraging the synergy among scalable architectural design and infrastructure efforts, we complete model training on more than 20 trillion tokens within 30 days, while achieving over 100 tokens per second (TPS) for inference at a cost of \$0.70 per million output tokens. To cultivate LongCat-Flash towards agentic intelligence, we conduct a large-scale pre-training on optimized mixtures, followed by targeted mid- and post-training on reasoning, code, and instructions, with further augmentation from synthetic data and tool use tasks. Comprehensive evaluations demonstrate that, as a non-thinking foundation model, LongCat-Flash delivers highly competitive performance among other leading models, with exceptional strengths in agentic tasks. The model checkpoint of LongCat-Flash is open-sourced to foster community research. LongCat Chat: https://longcat.ai Hugging Face: https://huggingface.co/meituan-longcat GitHub: https://github.com/meituan-longcat

cs.CL↗

arXiv2025

Ensembling Large Language Models for Code Vulnerability Detection: An Empirical Evaluation

Code vulnerability detection is crucial for ensuring the security and reliability of modern software systems. Recently, Large Language Models (LLMs) have shown promising capabilities in this domain. However, notable discrepancies in detection results often arise when analyzing identical code segments across different training stages of the same model or among architecturally distinct LLMs. While such inconsistencies may compromise detection stability, they also highlight a key opportunity: the latent complementarity among models can be harnessed through ensemble learning to create more robust vulnerability detection systems. In this study, we explore the potential of ensemble learning to enhance the performance of LLMs in source code vulnerability detection. We conduct comprehensive experiments involving five LLMs (i.e., DeepSeek-Coder-6.7B, CodeLlama-7B, CodeLlama-13B, CodeQwen1.5-7B, and StarCoder2-15B), using three ensemble strategies (i.e., Bagging, Boosting, and Stacking). These experiments are carried out across three widely adopted datasets (i.e., Devign, ReVeal, and BigVul). Inspired by Mixture of Experts (MoE) techniques, we further propose Dynamic Gated Stacking (DGS), a Stacking variant tailored for vulnerability detection. Our results demonstrate that ensemble approaches can significantly improve detection performance, with Boosting excelling in scenarios involving imbalanced datasets. Moreover, DGS consistently outperforms traditional Stacking, particularly in handling class imbalance and multi-class classification tasks. These findings offer valuable insights into building more reliable and effective LLM-based vulnerability detection systems through ensemble learning.

cs.SE↗

arXiv2025

VulAgent: Hypothesis-Validation based Multi-Agent Vulnerability Detection

The application of language models to project-level vulnerability detection remains challenging, owing to the dual requirement of accurately localizing security-sensitive code and correctly correlating and reasoning over complex program context. We present VulAgent, a multi-agent vulnerability detection framework based on hypothesis validation. Our design is inspired by how human auditors review code: when noticing a sensitive operation, they form a hypothesis about a possible vulnerability, consider potential trigger paths, and then verify the hypothesis against the surrounding context. VulAgent implements a semantics-sensitive, multi-view detection pipeline: specialized agents, each aligned to a specific analysis perspective (e.g., memory, authorization), collaboratively surface and precisely localize sensitive code sites with higher coverage. Building on this, VulAgent adopts a hypothesis-validation paradigm: for each vulnerability report, it builds hypothesis conditions and a trigger path, steering the LLM to target the relevant program context and defensive checks during verification, which reduces false positives. On average across the two datasets, VulAgent improves overall accuracy by 6.6%, increases the correct identification rate of vulnerable--fixed code pairs by up to 450% (246% on average), and reduces the false positive rate by about 36% compared with state-of-the-art LLM-based baselines.

cs.SE↗

arXiv2025

StackTrans: From Large Language Model to Large Pushdown Automata Model

The Transformer architecture has emerged as a landmark advancement within the broad field of artificial intelligence, effectively catalyzing the advent of large language models (LLMs). However, despite its remarkable capabilities and the substantial progress it has facilitated, the Transformer architecture still has some limitations. One such intrinsic limitation is its inability to effectively capture the Chomsky hierarchy, such as regular expressions or deterministic context-free grammars. Drawing inspiration from pushdown automata, which efficiently resolve deterministic context-free grammars using stacks, we propose StackTrans to address the aforementioned issue within LLMs. Unlike previous approaches that modify the attention computation, StackTrans explicitly incorporates hidden state stacks between Transformer layers. This design maintains compatibility with existing frameworks like flash-attention. Specifically, our design features stack operations -- such as pushing and popping hidden states -- that are differentiable and can be learned in an end-to-end manner. Our comprehensive evaluation spans benchmarks for both Chomsky hierarchies and large-scale natural languages. Across these diverse tasks, StackTrans consistently outperforms standard Transformer models and other baselines. We have successfully scaled StackTrans up from 360M to 7B parameters. In particular, our from-scratch pretrained model StackTrans-360M outperforms several larger open-source LLMs with 2-3x more parameters, showcasing its superior efficiency and reasoning capability.

cs.SE↗

arXiv2025

Molecular Dynamics Simulations of Nanoscale Friction on Illite Clay: Effects of Solvent Salt Ions and Electric Double Layer

Quick clay is a highly sensitive soil that transforms rapidly from solid to liquid under minor stress, as a result of long-term salt leaching that drastically reduces shear strength. Stabilizing it is both costly and carbon-intensive, significantly impacting construction emissions in regions like Norway. Developing greener stabilization methods is challenging due to limited understanding of the weakening mechanisms and the specific roles of different salts. In this study, we use molecular dynamics (MD) simulations to investigate the sliding behavior of illite platelets, the key component in Norwegian quick clay, and how it is affected by the different ions in the solution surrounding the surface. We examine the impact of monovalent (NaCl, KCl, CsCl) and divalent (MgCl2 and CaCl2) salts on platelet-surface interactions, focusing on the friction enhancement brought by divalent salts and how the electric double layer (EDL) structure mediates frictional behavior. We find that divalent cations sit higher on top of the surface, and lead to an increase in friction, while monovalent cations sit closer to the surface. By providing a detailed analysis of these interactions, the study offers a novel framework for understanding the role of salts in clay mechanics and highlights opportunities to design environmentally friendly stabilizers as alternatives to traditional lime and cement.

physics.geo-ph↗

arXiv2025

Stochasticity-aware No-Reference Point Cloud Quality Assessment

The evolution of point cloud processing algorithms necessitates an accurate assessment for their quality. Previous works consistently regard point cloud quality assessment (PCQA) as a MOS regression problem and devise a deterministic mapping, ignoring the stochasticity in generating MOS from subjective tests. This work presents the first probabilistic architecture for no-reference PCQA, motivated by the labeling process of existing datasets. The proposed method can model the quality judging stochasticity of subjects through a tailored conditional variational autoencoder (CVAE) and produces multiple intermediate quality ratings. These intermediate ratings simulate the judgments from different subjects and are then integrated into an accurate quality prediction, mimicking the generation process of a ground truth MOS. Specifically, our method incorporates a Prior Module, a Posterior Module, and a Quality Rating Generator, where the former two modules are introduced to model the judging stochasticity in subjective tests, while the latter is developed to generate diverse quality ratings. Extensive experiments indicate that our approach outperforms previous cutting-edge methods by a large margin and exhibits gratifying cross-dataset robustness. Codes are available at https://git.openi.org.cn/OpenPointCloud/nrpcqa.

cs.CV↗

arXiv2025

DIME:Diffusion-Based Maximum Entropy Reinforcement Learning

Maximum entropy reinforcement learning (MaxEnt-RL) has become the standard approach to RL due to its beneficial exploration properties. Traditionally, policies are parameterized using Gaussian distributions, which significantly limits their representational capacity. Diffusion-based policies offer a more expressive alternative, yet integrating them into MaxEnt-RL poses challenges-primarily due to the intractability of computing their marginal entropy. To overcome this, we propose Diffusion-Based Maximum Entropy RL (DIME). \emph{DIME} leverages recent advances in approximate inference with diffusion models to derive a lower bound on the maximum entropy objective. Additionally, we propose a policy iteration scheme that provably converges to the optimal diffusion policy. Our method enables the use of expressive diffusion-based policies while retaining the principled exploration benefits of MaxEnt-RL, significantly outperforming other diffusion-based methods on challenging high-dimensional control benchmarks. It is also competitive with state-of-the-art non-diffusion based RL methods while requiring fewer algorithmic design choices and smaller update-to-data ratios, reducing computational complexity.

cs.LG↗

arXiv2025

Computational Thinking Reasoning in Large Language Models

While large language models (LLMs) have demonstrated remarkable reasoning capabilities, they often struggle with complex tasks that require specific thinking paradigms, such as divide-and-conquer and procedural deduction, \etc Previous researches integrate external, reliable tools to alleviate logical inconsistencies and hallucinations in LLMs' problem-solving processes. However, we argue that the root challenge is more profound: LLMs lack the complex thinking paradigms (\ie, computational thinking) during reasoning. In this paper, we propose Computational Thinking Model (CTM), a novel framework that incorporates computational thinking paradigms into LLMs. This framework enables LLMs to reformulate complex problems through decomposition, abstraction, reduction, and simulation, among other techniques. Specifically, live code execution is seamlessly integrated into the reasoning process, allowing CTM to think by computing. CTM directly instills computational thinking objectives into LLMs through tailored reinforcement learning rewards, which encourages problem simplification, modular planning, and iterative verification. We conduct extensive evaluations on multiple code generation and mathematical benchmarks. The results demonstrate that CTM outperforms conventional reasoning models and tool-augmented baselines in terms of accuracy, interpretability, and generalizability. We hope this study offers valuable insights for AI reasoning, where LLMs can transform problems into robust, verifiable, and scalable computational workflows, much like computer scientists do.

cs.SE↗

arXiv2025

CodeDPO: Aligning Code Models with Self Generated and Verified Source Code

Code generation models have shown significant potential for programming tasks. However, existing training methods like supervised fine-tuning face key limitations: they do not effectively teach models to prioritize correct over incorrect solutions in ambiguous situations, nor do they effectively optimize the runtime efficiency of the generated code. To address these challenges, we propose CodeDPO, a framework that integrates preference learning into code generation to improve two key code preference factors: code correctness and efficiency. CodeDPO employs a novel dataset construction method, utilizing a self-generation-and-validation mechanism that simultaneously generates and evaluates code and test cases. The underlying assumption is that test cases executable by multiple code snippets provide more reliable validation, and code that passes more tests is more likely to be correct. Through this self-validation process, our PageRank-inspired algorithm iteratively updates the ranking score of each code snippet, ultimately creating a code preference optimization dataset based on correctness and efficiency. CodeDPO is flexible and scalable, generating diverse preference optimization data without depending on external resources. Through comprehensive evaluations of five widely used benchmarks, CodeDPO demonstrates significant improvements in correctness and efficiency compared to existing methods. Our experiments prove that CodeDPO enhances the capabilities of LLMs in code generation and provides a robust foundation for conducting code preference optimization in more complex and challenging real-world scenarios.

cs.SE↗

arXiv2025

Focused-DPO: Enhancing Code Generation Through Focused Preference Optimization on Error-Prone Points

Code generation models have shown significant potential for automating programming tasks. However, the challenge of generating accurate and reliable code persists due to the highly complex and long-reasoning nature of the task. Even state-of-the-art models often fail in code generation due to small errors, which can drastically affect the overall functionality of code. Our study identifies that current models tend to produce errors concentrated at specific error-prone points, which significantly impacts the accuracy of the generated code. To address this issue, we introduce Focused-DPO, a framework that enhances code generation by directing preference optimization towards these critical error-prone areas. This approach builds on Direct Preference Optimization, emphasizing accuracy in parts prone to errors. Additionally, we develop a method called Error-Point Identification, which constructs a dataset that targets these problematic points without requiring costly human annotations. Our experiments on benchmarks such as HumanEval(+), MBPP(+), and LiveCodeBench demonstrate that Focused-DPO significantly improves the precision and reliability of code generation, reducing common errors and enhancing overall code quality. By focusing on error-prone points, Focused-DPO advances the accuracy and functionality of model-generated code.

cs.SE↗

arXiv2025

LLM-Powered Test Case Generation for Detecting Bugs in Plausible Programs

Detecting tricky bugs in plausible programs, those that pass existing test suites yet still contain bugs, remains a significant challenge in software testing. To address this problem, we propose TrickCatcher, an LLM-powered approach to generating test cases for uncovering bugs in plausible programs. TrickCatcher operates in three stages: First, it uses an LLM to generate program variants based on the program under test (PUT) and its specification. Second, it employs an LLM to construct an input generator from the specification for producing test inputs. Finally, these inputs are executed on both the PUT and its program variants to detect inconsistencies in their outputs. We evaluate TrickCatcher on two datasets, TrickyBugs and EvalPlus, which include 366 human-written and 151 AI-generated plausible programs with tricky bugs. TrickCatcher achieves recall, precision, and F1 scores that are 1.80x, 2.65x, and 1.66x those of the state-of-the-art baselines, respectively. Code and data used are available at https://github.com/RinCloud/TrickCatcher.

cs.SE↗

arXiv2025

Theoretical Proof that Auto-regressive Language Models Collapse when Real-world Data is a Finite Set

Auto-regressive language models (LMs) have been widely used to generate data in data-scarce domains to train new LMs, compensating for the scarcity of real-world data. Previous work experimentally found that LMs collapse when trained on recursively generated data. This paper presents a theoretical proof: once a corpus (such as a subset of the World Wide Web) begins to incorporate generated data and no new real-world data is added to the corpus, then no matter how small the amount of data each LM generates and contributes to the corpus, LM collapse is inevitable after sufficient time. This finding suggests that attempts to mitigate collapse by limiting the quantity of synthetic data in the corpus are fundamentally insufficient. Instead, avoiding collapse hinges on ensuring the quality of synthetic data.

cs.CL↗