arXiv subjects

Ensheng Shi

Publications and source records attributed to Ensheng Shi.

At least 19 recordsLinked to original sources

arXiv2026

Adaptive Critical Token-Aware Retrieval for Repository-Level Code Generation

The repository-level code generation task requires synthesizing code that satisfies task requirements while remaining consistent with the target repository context. Since real-world repositories often exceed the input length limits of LLMs, existing approaches commonly adopt retrieval-augmented generation (RAG) to provide repository-specific context. Despite improving repository-context retrieval, existing methods typically provide context as task-level support, without explicitly identifying the critical tokens that require fine-grained repository context during generation. During the autoregressive generation process of LLMs, errors often concentrate at a small number of decisive positions: once such tokens are generated incorrectly, subsequent code may follow an incorrect semantic path and eventually lead to functional failure. We refer to these positions as "critical tokens". In this paper, we propose ACToR, an adaptive critical token-aware retrieval framework for repository-level code generation. ACToR identifies critical tokens during generation and triggers targeted retrieval on demand to provide repository context at these decisive positions. In addition, we design a position-aware weighting method for dense retrievers to prioritize context that is more informative for generation. We evaluate ACToR on two representative repository-level benchmarks, RepoExec and CoderEval. Experimental results show that ACToR consistently outperforms state-of-the-art methods, achieving relative improvements of 8.4% on RepoExec and 15.4% on CoderEval. Beyond performance gains, we systematically quantify the impact of critical tokens, revealing their central role in major generation failures and highlighting the necessity of targeted retrieval strategies. We provide the code and data at https://github.com/DeepSoftwareAnalytics/ACToR.

cs.SE↗

arXiv2026

Efficient SWE Agent Benchmarking via Trajectory-Aware Evaluation

Evaluating software engineering agents on realistic benchmarks is costly, since each task may require multi-step code exploration, modification, and test execution. Existing efficient evaluation methods select representative subsets to estimate full-benchmark performance, but are largely result-only: they fit historical pass/fail response matrices or static task semantics, discarding how agents solve problems. We propose PTA-IRT, a Privileged Trajectory-Aware Item Response Theory framework that fuses process and outcome signals. Historical execution trajectories supply process-level evidence beyond pass/fail, such as explored context, attempted edits, and solving paths, which PTA-IRT uses as privileged information for calibration subset selection and ability estimation. Under low calibration budgets, PTA-IRT consistently outperforms prior IRT baselines on score and ranking recovery across four SWE benchmarks. Code and data are publicly available at https://github.com/DeepSoftwareAnalytics/PTA-IRT.

cs.SE↗

arXiv2026

SWE-Prime: Fewer Trajectories, Better Performance

To improve large language models' ability to resolve real-world software issues, prior work has focused on constructing large-scale agent trajectory datasets and performing supervised fine-tuning (SFT) on successful trajectories. However, task success does not guarantee high-quality supervision: successful trajectories may still contain ineffective, redundant, or risky steps. Directly using such trajectories for SFT can introduce noisy supervision and encourage models to imitate undesirable problem-solving behaviors. Therefore, we propose SWE-Prime, a multi-granularity, two-stage SFT data selection method that progressively filters training data at the trajectory and segment levels. Specifically, the first stage performs trajectory-level screening based on process quality, result quality, and data representativeness, selecting a high-quality and representative subset of successful trajectories. The second stage performs segment-level selection by grouping consecutive steps into semantic segments and assessing each segment based on its contribution to the final solution, learnability, and potential risks. During SFT, all segments remain in the sequence to preserve context, while only selected segments contribute to the loss computation. Experiments on SWE-Bench Pro and SWE-Bench Verified show that training on the 10% trajectory subset selected by SWE-Prime outperforms training on the full resolved dataset, yielding relative performance gains of up to 12.2% and 24.2%, respectively.

cs.SE↗

arXiv2026

PhoenixRepair: Rethinking Repair Strategy Exploration in Software Agents

While Large Language Models have greatly advanced automated issue resolution, existing agent-based methods exhibit a fundamental limitation in their insufficient exploration of repair strategies. This insufficiency manifests in two key aspects. First, the exploration of multiple potential edit locations is limited. Second, the exploration of repair attempts at each location is also insufficient. To address these challenges, we present PhoenixRepair, a multi-agent framework that systematically explores multiple candidate edit locations and performs iterative reflection and refinement on patch generation, thereby expanding the search space of repair strategies. Our framework begins with multi-location sampling, optionally augmented with graph-based localization information for difficult tasks, followed by iterative reflection and refinement to generate better patches, culminating in final-round generation guided by distilled insights from all historical attempts. Experiments on SWE-bench-Verified demonstrate that PhoenixRepair achieves the largest relative improvement of 7.8\% over SWE-agent under DeepSeek-V3.1, and attains the highest resolved rate of 76.0\% Pass@1 under MiniMax-M2.5. Meanwhile, it achieves higher fault localization accuracy than existing approaches. Our code is available at https://github.com/DeepSoftwareAnalytics/PhoenixRepair.

cs.AI↗

arXiv2026

Yet Even Less Is Even Better For Agentic, Reasoning, and Coding LLMs

Training effective software engineering agents requires large volumes of task-specific trajectories, incurring substantial data construction costs. Inspired by the "Less-Is-More" hypothesis in mathematical reasoning, we investigate its extension to agentic scenarios and propose an end-to-end training framework that achieves superior agentic capabilities with fewer but higher-quality training trajectories. This is achieved via STITCH (Sliding-memory Trajectory Inference and Task Chunking Heuristic), a coarse-to-fine mechanism that filters low-value noise and retains decision-critical tokens to maximize training signal quality. We conduct experiments across multiple agent frameworks (e.g., mini-SWE-agent, MSWE-agent), model scales (30B to 355B), and multilingual settings (Python, Java, and ArkTS). On SWE-bench Verified, models trained with STITCH achieve up to 63.16% relative improvement over base models. On Multi-SWE-bench (Java), MiniMax-M2.5-STITCH achieves 43.75% with our CodeArts Agent scaffold (+16.67%). On HarmonyOS (ArkTS), GLM-4.7-STITCH improves the compilation pass rate to 61.31% (+43.34%) with less than 1K training trajectories. Our results confirm that the "Less-Is-More" paradigm generalizes effectively to complex agentic tasks across diverse languages and model scales.

cs.SE↗

arXiv2026

Towards an Understanding of Context Utilization in Code Intelligence

Code intelligence is an emerging domain in software engineering, aiming to improve the effectiveness and efficiency of various code-related tasks. Recent research suggests that incorporating contextual information beyond the basic original task inputs (i.e., source code) can substantially enhance model performance. Such contextual signals may be obtained directly or indirectly from sources such as API documentation or intermediate representations like abstract syntax trees can significantly improve the effectiveness of code intelligence. Despite growing academic interest, there is a lack of systematic analysis of context in code intelligence. To address this gap, we conduct an extensive literature review of 146 relevant studies published between September 2007 and August 2024. Our investigation yields four main contributions. (1) A quantitative analysis of the research landscape, including publication trends, venues, and the explored domains; (2) A novel taxonomy of context types used in code intelligence; (3) A task-oriented analysis investigating context integration strategies across diverse code intelligence tasks; (4) A critical evaluation of evaluation methodologies for context-aware methods. Based on these findings, we identify fundamental challenges in context utilization in current code intelligence systems and propose a research roadmap that outlines key opportunities for future research.

cs.SE↗

arXiv2026

DRAINCODE: Stealthy Energy Consumption Attacks on Retrieval-Augmented Code Generation via Context Poisoning

Large language models (LLMs) have demonstrated impressive capabilities in code generation by leveraging retrieval-augmented generation (RAG) methods. However, the computational costs associated with LLM inference, particularly in terms of latency and energy consumption, have received limited attention in the security context. This paper introduces DrainCode, the first adversarial attack targeting the computational efficiency of RAG-based code generation systems. By strategically poisoning retrieval contexts through a mutation-based approach, DrainCode forces LLMs to produce significantly longer outputs, thereby increasing GPU latency and energy consumption. We evaluate the effectiveness of DrainCode across multiple models. Our experiments show that DrainCode achieves up to an 85% increase in latency, a 49% increase in energy consumption, and more than a 3x increase in output length compared to the baseline. Furthermore, we demonstrate the generalizability of the attack across different prompting strategies and its effectiveness compared to different defenses. The results highlight DrainCode as a potential method for increasing the computational overhead of LLMs, making it useful for evaluating LLM security in resource-constrained environments. We provide code and data at https://github.com/DeepSoftwareAnalytics/DrainCode.

cs.SE↗

arXiv2026

AlignCoder: Aligning Retrieval with Target Intent for Repository-Level Code Completion

Repository-level code completion remains a challenging task for existing code large language models (code LLMs) due to their limited understanding of repository-specific context and domain knowledge. While retrieval-augmented generation (RAG) approaches have shown promise by retrieving relevant code snippets as cross-file context, they suffer from two fundamental problems: misalignment between the query and the target code in the retrieval process, and the inability of existing retrieval methods to effectively utilize the inference information. To address these challenges, we propose AlignCoder, a repository-level code completion framework that introduces a query enhancement mechanism and a reinforcement learning based retriever training method. Our approach generates multiple candidate completions to construct an enhanced query that bridges the semantic gap between the initial query and the target code. Additionally, we employ reinforcement learning to train an AlignRetriever that learns to leverage inference information in the enhanced query for more accurate retrieval. We evaluate AlignCoder on two widely-used benchmarks (CrossCodeEval and RepoEval) across five backbone code LLMs, demonstrating an 18.1% improvement in EM score compared to baselines on the CrossCodeEval benchmark. The results show that our framework achieves superior performance and exhibits high generalizability across various code LLMs and programming languages.

cs.SE↗

arXiv2026

ShortCoder: Knowledge-Augmented Syntax Optimization for Token-Efficient Code Generation

Code generation tasks aim to automate the conversion of user requirements into executable code, significantly reducing manual development efforts and enhancing software productivity. The emergence of large language models (LLMs) has significantly advanced code generation, though their efficiency is still impacted by certain inherent architectural constraints. Each token generation necessitates a complete inference pass, requiring persistent retention of contextual information in memory and escalating resource consumption. While existing research prioritizes inference-phase optimizations such as prompt compression and model quantization, the generation phase remains underexplored. To tackle these challenges, we propose a knowledge-infused framework named ShortCoder, which optimizes code generation efficiency while preserving semantic equivalence and readability. In particular, we introduce: (1) ten syntax-level simplification rules for Python, derived from AST-preserving transformations, achieving 18.1% token reduction without functional compromise; (2) a hybrid data synthesis pipeline integrating rule-based rewriting with LLM-guided refinement, producing ShorterCodeBench, a corpus of validated tuples of original code and simplified code with semantic consistency; (3) a fine-tuning strategy that injects conciseness awareness into the base LLMs. Extensive experimental results demonstrate that ShortCoder consistently outperforms state-of-the-art methods on HumanEval, achieving an improvement of 18.1%-37.8% in generation efficiency over previous methods while ensuring the performance of code generation.

cs.SE↗

arXiv2025

UCoder: Unsupervised Code Generation by Internal Probing of Large Language Models

Large language models (LLMs) have demonstrated remarkable capabilities in code generation tasks. However, their effectiveness heavily relies on supervised training with extensive labeled (e.g., question-answering pairs) or unlabeled datasets (e.g., code snippets), which are often expensive and difficult to obtain at scale. To address this limitation, this paper introduces a method IPC, an unsupervised framework that leverages Internal Probing of LLMs for Code generation without any external corpus, even unlabeled code snippets. We introduce the problem space probing, test understanding probing, solution space probing, and knowledge consolidation and reinforcement to probe the internal knowledge and confidence patterns existing in LLMs. Further, IPC identifies reliable code candidates through self-consistency mechanisms and representation-based quality estimation to train UCoder (coder with unsupervised learning). We validate the proposed approach across multiple code benchmarks, demonstrating that unsupervised methods can achieve competitive performance compared to supervised approaches while significantly reducing the dependency on labeled data and computational resources. Analytic experiments reveal that internal model states contain rich signals about code quality and correctness, and that properly harnessing these signals enables effective unsupervised learning for code generation tasks, opening new directions for training code LLMs in resource-constrained scenarios.

cs.CL↗

arXiv2025

SimpleDevQA: Benchmarking Large Language Models on Development Knowledge QA

The Development Knowledge Question Answering (Dev Knowledge QA) task aims to provide natural language answers to knowledge-seeking questions during software development. To investigate its importance and to what extent it has been explored, we analyze real user-LLM dialogues from WildChat and find that: (1) The Dev Knowledge QA task accounts for 39.6% of interactions(highest among all tasks), revealing broad knowledge needs beyond code generation (32.3%). (2) Only 27.5% of real Dev Knowledge QA dialogues focus on code understanding, leaving out development knowledge-seeking. (3) Only 17.1% of real-world Dev Knowledge QA dialogues can be used for constructing a benchmark. Existing benchmarks have two primary limitations for evaluating the Dev Knowledge QA capability of LLMs. First, existing benchmarks offer a limited development knowledge scope, mainly focusing on code understanding and neglecting broader knowledge during development. Second, some benchmarks are not built from real user queries. To bridge this gap, we design a three-phase pipeline that transforms real-world dialogue into simple development knowledge-seeking QA pairs. Through this pipeline, we introduce SimpleDevQA, a multilingual benchmark derived from real user dialogues. It contains 2,740 QA pairs in three languages (English, Chinese, and Russian), and focuses on questions with unique, short, and verifiable answers for accurate and simple evaluation. Experiments show that: Code LLMs generally outperform general LLMs of similar scale; Knowledge injection with the Retrieval-Augmented Generation (RAG) strategy can boost LLM accuracy by 11.3% on average; LLMs show systematic overconfidence in Dev Knowledge QA, and the answering accuracy of LLMs shows a positive correlation with their stated confidence; Generally, LLMs with stronger code generation performance also exhibit stronger performance in Dev Knowledge QA.

cs.SE↗

arXiv2025

From Code Foundation Models to Agents and Applications: A Comprehensive Survey and Practical Guide to Code Intelligence

Large language models (LLMs) have fundamentally transformed automated software development by enabling direct translation of natural language descriptions into functional code, driving commercial adoption through tools like Github Copilot (Microsoft), Cursor (Anysphere), Trae (ByteDance), and Claude Code (Anthropic). While the field has evolved dramatically from rule-based systems to Transformer-based architectures, achieving performance improvements from single-digit to over 95\% success rates on benchmarks like HumanEval. In this work, we provide a comprehensive synthesis and practical guide (a series of analytic and probing experiments) about code LLMs, systematically examining the complete model life cycle from data curation to post-training through advanced prompting paradigms, code pre-training, supervised fine-tuning, reinforcement learning, and autonomous coding agents. We analyze the code capability of the general LLMs (GPT-4, Claude, LLaMA) and code-specialized LLMs (StarCoder, Code LLaMA, DeepSeek-Coder, and QwenCoder), critically examining the techniques, design decisions, and trade-offs. Further, we articulate the research-practice gap between academic research (e.g., benchmarks and tasks) and real-world deployment (e.g., software-related code tasks), including code correctness, security, contextual awareness of large codebases, and integration with development workflows, and map promising research directions to practical needs. Last, we conduct a series of experiments to provide a comprehensive analysis of code pre-training, supervised fine-tuning, and reinforcement learning, covering scaling law, framework selection, hyperparameter sensitivity, model architectures, and dataset comparisons.

cs.SE↗

arXiv2025

EffiReasonTrans: RL-Optimized Reasoning for Code Translation

Code translation is a crucial task in software development and maintenance. While recent advancements in large language models (LLMs) have improved automated code translation accuracy, these gains often come at the cost of increased inference latency, hindering real-world development workflows that involve human-in-the-loop inspection. To address this trade-off, we propose EffiReasonTrans, a training framework designed to improve translation accuracy while balancing inference latency. We first construct a high-quality reasoning-augmented dataset by prompting a stronger language model, DeepSeek-R1, to generate intermediate reasoning and target translations. Each (source code, reasoning, target code) triplet undergoes automated syntax and functionality checks to ensure reliability. Based on this dataset, we employ a two-stage training strategy: supervised fine-tuning on reasoning-augmented samples, followed by reinforcement learning to further enhance accuracy and balance inference latency. We evaluate EffiReasonTrans on six translation pairs. Experimental results show that it consistently improves translation accuracy (up to +49.2% CA and +27.8% CodeBLEU compared to the base model) while reducing the number of generated tokens (up to -19.3%) and lowering inference latency in most cases (up to -29.0%). Ablation studies further confirm the complementary benefits of the two-stage training framework. Additionally, EffiReasonTrans demonstrates improved translation accuracy when integrated into agent-based frameworks. Our code and data are available at https://github.com/DeepSoftwareAnalytics/EffiReasonTrans.

cs.SE↗

arXiv2025

HumanEvo: An Evolution-aware Benchmark for More Realistic Evaluation of Repository-level Code Generation

To evaluate the repository-level code generation capabilities of Large Language Models (LLMs) in complex real-world software development scenarios, many evaluation methods have been developed. These methods typically leverage contextual code from the latest version of a project to assist LLMs in accurately generating the desired function. However, such evaluation methods fail to consider the dynamic evolution of software projects over time, which we refer to as evolution-ignored settings. This in turn results in inaccurate evaluation of LLMs' performance. In this paper, we conduct an empirical study to deeply understand LLMs' code generation performance within settings that reflect the evolution nature of software development. To achieve this, we first construct an evolution-aware repository-level code generation dataset, namely HumanEvo, equipped with an automated execution-based evaluation tool. Second, we manually categorize HumanEvo according to dependency levels to more comprehensively analyze the model's performance in generating functions with different dependency levels. Third, we conduct extensive experiments on HumanEvo with seven representative and diverse LLMs to verify the effectiveness of the proposed benchmark. We obtain several important findings through our experimental study. For example, we find that previous evolution-ignored evaluation methods result in inflated performance of LLMs, with performance overestimations ranging from 10.0% to 61.1% under different context acquisition methods, compared to the evolution-aware evaluation approach. Based on the findings, we give actionable suggestions for more realistic evaluation of LLMs on code generation. We also build a shared evolution-aware code generation toolbox to facilitate future research.

cs.SE↗

arXiv2025

Top General Performance = Top Domain Performance? DomainCodeBench: A Multi-domain Code Generation Benchmark

With the rapid advancement of large language models (LLMs), extensive research has been conducted to investigate the code generation capabilities of LLMs. However, existing efforts primarily focus on general-domain tasks, leaving LLMs' code generation performance in real-world application domains underexplored. This raises a critical question: can a model's general-domain coding ability reliably represent its ability in specialized domains? In this paper, we introduce DomainCodeBench, a multi-domain code generation benchmark designed to systematically evaluate LLMs across 12 software application domains and 15 programming languages. DomainCodeBench contains 2,400 manually verified tasks with ground truth, human-annotated docstrings, and fine-grained dependency information to ensure more coverage of domain-specific challenges. Specifically, we first identify the most popular application domains by topic mining. Then, we curate coding tasks based on commonly used frameworks and platforms in each domain. We obtain several findings through extensive experiments on DomainCodeBench with ten mainstream LLMs. (1) Performance decoupling: experiments reveal that top general-domain models do not consistently excel in specific application domains; (2) Domain-specific weaknesses: LLMs often fail due to domain knowledge gaps and third-party library misusage; (3) Contextual enhancement: we show that augmenting prompts with domain-specific knowledge improves performance by around 38.17%, providing actionable insights for performance optimization. Our replication package, including the benchmark, source code, and experimental results, is available at https://github.com/DeepSoftwareAnalytics/DomainCodeBench.

cs.SE↗

arXiv2024

SECRET: Towards Scalable and Efficient Code Retrieval via Segmented Deep Hashing

Code retrieval, which retrieves code snippets based on users' natural language descriptions, is widely used by developers and plays a pivotal role in real-world software development. The advent of deep learning has shifted the retrieval paradigm from lexical-based matching towards leveraging deep learning models to encode source code and queries into vector representations, facilitating code retrieval according to vector similarity. Despite the effectiveness of these models, managing large-scale code database presents significant challenges. Previous research proposes deep hashing-based methods, which generate hash codes for queries and code snippets and use Hamming distance for rapid recall of code candidates. However, this approach's reliance on linear scanning of the entire code base limits its scalability. To further improve the efficiency of large-scale code retrieval, we propose a novel approach SECRET (Scalable and Efficient Code Retrieval via SegmEnTed deep hashing). SECRET converts long hash codes calculated by existing deep hashing approaches into several short hash code segments through an iterative training strategy. After training, SECRET recalls code candidates by looking up the hash tables for each segment, the time complexity of recall can thus be greatly reduced. Extensive experimental results demonstrate that SECRET can drastically reduce the retrieval time by at least 95% while achieving comparable or even higher performance of existing deep hashing approaches. Besides, SECRET also exhibits superior performance and efficiency compared to the classical hash table-based approach known as LSH under the same number of hash tables.

cs.SE↗

arXiv2024

Agents in Software Engineering: Survey, Landscape, and Vision

In recent years, Large Language Models (LLMs) have achieved remarkable success and have been widely used in various downstream tasks, especially in the tasks of the software engineering (SE) field. We find that many studies combining LLMs with SE have employed the concept of agents either explicitly or implicitly. However, there is a lack of an in-depth survey to sort out the development context of existing works, analyze how existing works combine the LLM-based agent technologies to optimize various tasks, and clarify the framework of LLM-based agents in SE. In this paper, we conduct the first survey of the studies on combining LLM-based agents with SE and present a framework of LLM-based agents in SE which includes three key modules: perception, memory, and action. We also summarize the current challenges in combining the two fields and propose future opportunities in response to existing challenges. We maintain a GitHub repository of the related papers at: https://github.com/DeepSoftwareAnalytics/Awesome-Agent4SE.

cs.SE↗

arXiv2024

You Augment Me: Exploring ChatGPT-based Data Augmentation for Semantic Code Search

Code search plays a crucial role in software development, enabling developers to retrieve and reuse code using natural language queries. While the performance of code search models improves with an increase in high-quality data, obtaining such data can be challenging and expensive. Recently, large language models (LLMs) such as ChatGPT have made remarkable progress in both natural and programming language understanding and generation, offering user-friendly interaction via simple prompts. Inspired by these advancements, we propose a novel approach ChatDANCE, which utilizes high-quality and diverse augmented data generated by a large language model and leverages a filtering mechanism to eliminate low-quality augmentations. Specifically, we first propose a set of ChatGPT prompting rules that are specifically designed for source code and queries. Then, we leverage ChatGPT to rewrite code and queries based on the according prompts and then propose a filtering mechanism which trains a cross-encoder from the backbone model UniXcoder to filter out code and query pairs with low matching scores. Finally, we re-train the backbone model using the obtained high-quality augmented data. Experimental results show that ChatDANCE achieves state-of-the-art performance, improving the best baseline by 13.2% (R@1) and 7% (MRR). Surprisingly, we find that this augment-filter-retrain strategy enables the backbone model (UniXcoder) to self-grow. Moreover, extensive experiments show the effectiveness of each component and ChatDANCE has stable performance under different hyperparameter settings. In addition, we conduct qualitative and quantitative analyses to investigate why ChatDANCE works well and find that it learns a more uniform distribution of representations and effectively aligns the code and query spaces.

cs.SE↗