In recent years, code intelligence has gained increasing importance in the field of automated software engineering. Meanwhile, the widespread adoption of Pretrained Language Models (PLMs) and Large Language Models (LLMs) has raised concerns regarding data contamination and its potential impact on model performance evaluation. This paper presents a systematic empirical study to investigate the fine-grained data contamination on code intelligence tasks. Our study involves diverse representative PLMs, namely RoBERTa and GPT-2, and LLMs, namely LLaMA and StarCoder, covering three major tasks: code translation, code generation, and code summarization. We categorize contamination scenarios into four types according to the code intelligence practice, namely input-only, output-only, unpaired, and paired contamination settings, and construct corresponding experimental and control groups for exploration. Experimental results show that, under the pre-training, fine-tuning, and inference paradigm adopted by PLMs, even deliberately injecting paired contamination does not lead to significant performance overestimation. But direct inference or small-scale fine-tuning uncovers the contamination effects. In contrast, LLMs with pre-training and inference paradigm are significantly affected by the paired contamination. Apart from the above, other contamination scenarios have no impact on both PLMs and LLMs. Our findings challenge the conventional belief that contamination inevitably leads to performance overestimation, providing new insights into the evaluation and deployment of code intelligence models.
Large Language Models (LLMs) and other automated techniques have been increasingly used to support software developers by generating software artifacts such as code snippets, patches, and comments. However, accurately assessing the correctness of these generated artifacts remains a significant challenge. On one hand, human evaluation provides high accuracy but is labor-intensive and lacks scalability. On the other hand, other existing automatic evaluation metrics are scalable and require minimal human effort, but they often fail to accurately reflect the actual correctness of generated software artifacts. In this paper, we present SWE-Judge, the first evaluation metric for LLM-as-Ensemble-Judge specifically designed to accurately assess the correctness of generated software artifacts. SWE-Judge first defines five distinct evaluation strategies, each implemented as an independent judge. A dynamic team selection mechanism then identifies the most appropriate subset of judges to produce a final correctness score through ensembling. We evaluate SWE-Judge across a diverse set of software engineering (SE) benchmarks, including CoNaLa, Card2Code, HumanEval-X, APPS, APR-Assess, and Summary-Assess. These benchmarks span three SE tasks: code generation, automated program repair, and code summarization. Experimental results demonstrate that SWE-Judge consistently achieves a higher correlation with human judgments, with improvements ranging from 5.9% to 183.8% over existing automatic metrics. Furthermore, SWE-Judge reaches agreement levels with human annotators that are comparable to inter-annotator agreement in code generation and program repair tasks. These findings underscore SWE-Judge's potential as a scalable and reliable alternative to human evaluation.
Large Language Models for code often entail significant computational complexity, which grows significantly with the length of the input code sequence. We propose LeanCode for code simplification to reduce training and prediction time, leveraging code contexts in utilizing attention scores to represent the tokens' importance. We advocate for the selective removal of tokens based on the average context-aware attention scores rather than average scores across all inputs. LeanCode uses the attention scores of `CLS' tokens within the encoder for classification tasks, such as code search. It also employs the encoder-decoder attention scores to determine token significance for sequence-to-sequence tasks like code summarization. Our evaluation shows LeanCode's superiority over the SOTAs DietCode and Slimcode, with improvements of 60% and 16% for code search, and 29% and 27% for code summarization, respectively.
Assessing the programming capabilities of Large Language Models (LLMs) is crucial for their effective use in software engineering. Current evaluations, however, predominantly measure the accuracy of generated code on static benchmarks, neglecting the critical aspect of model robustness during programming tasks. While adversarial attacks offer insights on model robustness, their effectiveness is limited and evaluation could be constrained. Current adversarial attack methods for robustness evaluation yield inconsistent results, struggling to provide a unified evaluation across different LLMs. We introduce EVALOOP, a novel assessment framework that evaluate the robustness from a self-consistency perspective, i.e., leveraging the natural duality inherent in popular software engineering tasks, e.g., code generation and code summarization. EVALOOP initiates a self-contained feedback loop: an LLM generates output (e.g., code) from an input (e.g., natural language specification), and then use the generated output as the input to produce a new output (e.g., summarizes that code into a new specification). EVALOOP repeats the process to assess the effectiveness of EVALOOP in each loop. This cyclical strategy intrinsically evaluates robustness without rely on any external attack setups, providing a unified metric to evaluate LLMs' robustness in programming. We evaluate 16 prominent LLMs (e.g., GPT-4.1, O4-mini) on EVALOOP and found that EVALOOP typically induces a 5.01%-19.31% absolute drop in pass@1 performance within ten loops. Intriguingly, robustness does not always align with initial performance (i.e., one-time query); for instance, GPT-3.5-Turbo, despite superior initial code generation compared to DeepSeek-V2, demonstrated lower robustness over repeated evaluation loop.
Recent advancements in source code summarization have leveraged transformer-based pre-trained models, including Large Language Models of Code (LLMCs), to automate and improve the generation of code summaries. However, existing methods often focus on generating a single high-quality summary for a given source code, neglecting scenarios where the generated summary might be inadequate and alternative options are needed. In this paper, we introduce Variational Prefix Tuning (VPT), a novel approach that enhances pre-trained models' ability to generate diverse yet accurate sets of summaries, allowing the user to choose the most suitable one for the given source code. Our method integrates a Conditional Variational Autoencoder (CVAE) framework as a modular component into pre-trained models, enabling us to model the distribution of observed target summaries and sample continuous embeddings to be used as prefixes to steer the generation of diverse outputs during decoding. Importantly, we construct our method in a parameter-efficient manner, eliminating the need for expensive model retraining, especially when using LLMCs. Furthermore, we employ a bi-criteria reranking method to select a subset of generated summaries, optimizing both the diversity and the accuracy of the options presented to users. We present extensive experimental evaluations using widely used datasets and current state-of-the-art pre-trained code summarization models to demonstrate the effectiveness of our approach and its adaptability across models.
Pre-trained code models rely heavily on high-quality pre-training data, particularly human-written reference comments that bridge code and natural language. However, these comments often become outdated as software evolves, degrading model performance. Large language models (LLMs) excel at generating high-quality code comments. We investigate whether replacing human-written comments with LLM-generated ones improves pre-training datasets. Since standard metrics cannot assess reference comment quality, we propose two novel reference-free evaluation tasks: code-comment inconsistency detection and semantic code search. Results show that LLM-generated comments are more semantically consistent with code than human-written ones, as confirmed by manual evaluation. Leveraging this finding, we rebuild the CodeSearchNet dataset with LLM-generated comments and re-pre-train CodeT5. Evaluations demonstrate that models trained on LLM-enhanced data outperform those using original human comments in code summarization, generation, and translation tasks. This work validates rebuilding pre-training datasets with LLMs to advance code intelligence, challenging the traditional reliance on human reference comments.
Understanding and navigating large-scale codebases remains a significant challenge in software engineering. Existing methods often treat code as flat text or focus primarily on local structural relationships, limiting their ability to provide holistic, context-aware information retrieval. We present Hierarchical Code Graph Summarization (HCGS), a novel approach that constructs a multi-layered representation of a codebase by generating structured summaries in a bottom-up fashion from a code graph. HCGS leverages the Language Server Protocol for language-agnostic code analysis and employs a parallel level-based algorithm for efficient summary generation. Through extensive evaluation on five diverse codebases totaling 7,531 functions, HCGS demonstrates significant improvements in code retrieval accuracy, achieving up to 82 percentage relative improvement in top-1 retrieval precision for large codebases like libsignal (27.15 percentage points), and perfect Pass@3 scores for smaller repositories. The system's hierarchical approach consistently outperforms traditional code-only retrieval across all metrics, with particularly substantial gains in larger, more complex codebases where understanding function relationships is crucial.
Human attention provides valuable yet underexploited signals for code LLM training, offering a perspective beyond purely machine-driven attention. Despite the complexity and cost of collecting eye-tracking data, there has also been limited progress in systematically using these signals for code LLM training. To address both issues, we propose a cohesive pipeline spanning augmentation and reward-based fine-tuning. Specifically, we introduce (1) an eye-tracking path augmentation method to expand programmer attention datasets, (2) a pattern abstraction step that refines raw fixations into learnable attention motifs, and (3) a reward-guided strategy for integrating these insights directly into a CodeT5 supervised fine-tuning process. Our experiments yield +7.16 in CodeBLEU on the CodeXGlue benchmark for code summarization, underscoring how uniting human and machine attention can boost code intelligence. We hope this work encourages broader exploration of human-centric methods in next-generation AI4SE.
Commenting code is a crucial activity in software development, as it aids in facilitating future maintenance and updates. To enhance the efficiency of writing comments and reduce developers' workload, researchers has proposed various automated code summarization (ACS) techniques to automatically generate comments/summaries for given code units. However, these ACS techniques primarily focus on generating summaries for code units at the method level. There is a significant lack of research on summarizing higher-level code units, such as file-level and module-level code units, despite the fact that summaries of these higher-level code units are highly useful for quickly gaining a macro-level understanding of software components and architecture. To fill this gap, in this paper, we conduct a systematic study on how to use LLMs for commenting higher-level code units, including file level and module level. These higher-level units are significantly larger than method-level ones, which poses challenges in handling long code inputs within LLM constraints and maintaining efficiency. To address these issues, we explore various summarization strategies for ACS of higher-level code units, which can be divided into three types: full code summarization, reduced code summarization, and hierarchical code summarization. The experimental results suggest that for summarizing file-level code units, using the full code is the most effective approach, with reduced code serving as a cost-efficient alternative. However, for summarizing module-level code units, hierarchical code summarization becomes the most promising strategy. In addition, inspired by the research on method-level ACS, we also investigate using the LLM as an evaluator to evaluate the quality of summaries of higher-level code units. The experimental results demonstrate that the LLM's evaluation results strongly correlate with human evaluations.
Code summarization is a critical task in natural language processing and software engineering, which aims to generate concise descriptions of source code. Recent advancements have improved the quality of these summaries, enhancing code readability and maintainability. However, the content of a repository or a class has not been considered in function code summarization. This study investigated the effectiveness of code summarization models beyond the function level, exploring the impact of class and repository contexts on the summary quality. The study involved revising benchmarks for evaluating models at class and repository levels, assessing baseline models, and evaluating LLMs with in-context learning to determine the enhancement of summary quality with additional context. The findings revealed that the fine-tuned state-of-the-art CodeT5+ base model excelled in code summarization, while incorporating few-shot learning and retrieved code chunks from RAG significantly enhanced the performance of LLMs in this task. Notably, the Deepseek Coder 1.3B and Starcoder2 15B models demonstrated substantial improvements in metrics such as BLEURT, METEOR, and BLEU-4 at both class and repository levels. Repository-level summarization exhibited promising potential but necessitates significant computational resources and gains from the inclusion of structured context. Lastly, we employed the recent SIDE code summarization metric in our evaluation. This study contributes to refining strategies for prompt engineering, few-shot learning, and RAG, addressing gaps in benchmarks for code summarization at various levels. Finally, we publish all study details, code, datasets, and results of evaluation in the GitHub repository available at https://github.com/kilimanj4r0/code-summarization-beyond-function-level.
Decoding strategies significantly influence the quality and diversity of the generated texts in large language models (LLMs), yet their impact on computational resource consumption, particularly GPU energy usage, is insufficiently studied. This paper investigates the relationship between text generation decoding methods and energy efficiency, focusing on the trade-off between generation quality and GPU energy consumption across diverse tasks and decoding configurations. By benchmarking multiple strategies across different text generation tasks, such as Translation, Code Summarization, and Math Problem Solving, we reveal how selecting appropriate decoding techniques with their tuned hyperparameters affects text quality and has measurable implications for resource utilization, emphasizing the need for balanced optimization. To the best of our knowledge, this study is among the first to explore decoding strategies in LLMs through the lens of energy consumption, offering actionable insights for designing resource-aware applications that maintain high-quality text generation.
Intermediate Representations (IRs) are essential in compiler design and program analysis, yet their comprehension by Large Language Models (LLMs) remains underexplored. This paper presents a pioneering empirical study to investigate the capabilities of LLMs, including GPT-4, GPT-3, Gemma 2, LLaMA 3.1, and Code Llama, in understanding IRs. We analyze their performance across four tasks: Control Flow Graph (CFG) reconstruction, decompilation, code summarization, and execution reasoning. Our results indicate that while LLMs demonstrate competence in parsing IR syntax and recognizing high-level structures, they struggle with control flow reasoning, execution semantics, and loop handling. Specifically, they often misinterpret branching instructions, omit critical IR operations, and rely on heuristic-based reasoning, leading to errors in CFG reconstruction, IR decompilation, and execution reasoning. The study underscores the necessity for IR-specific enhancements in LLMs, recommending fine-tuning on structured IR datasets and integration of explicit control flow models to augment their comprehension and handling of IR-related tasks.
Recently, large language models (LLMs) have been deployed to tackle various software engineering (SE) tasks like code generation, significantly advancing the automation of SE tasks. However, assessing the quality of these LLM-generated code and text remains challenging. The commonly used Pass@k metric necessitates extensive unit tests and configured environments, demands a high labor cost, and is not suitable for evaluating LLM-generated text. Conventional metrics like BLEU, which measure only lexical rather than semantic similarity, have also come under scrutiny. In response, a new trend has emerged to employ LLMs for automated evaluation, known as LLM-as-a-judge. These LLM-as-a-judge methods are claimed to better mimic human assessment than conventional metrics without relying on high-quality reference answers. Nevertheless, their exact human alignment in SE tasks remains unexplored. In this paper, we empirically explore LLM-as-a-judge methods for evaluating SE tasks, focusing on their alignment with human judgments. We select seven LLM-as-a-judge methods that utilize general-purpose LLMs, alongside two LLMs specifically fine-tuned for evaluation. After generating and manually scoring LLM responses on three recent SE datasets of code translation, code generation, and code summarization, we then prompt these methods to evaluate each response. Finally, we compare the scores generated by these methods with human evaluation. The results indicate that output-based methods reach the highest Pearson correlation of 81.32 and 68.51 with human scores in code translation and generation, achieving near-human evaluation, noticeably outperforming ChrF++, one of the best conventional metrics, at 34.23 and 64.92. Such output-based methods prompt LLMs to output judgments directly, and exhibit more balanced score distributions that resemble human score patterns. Finally, we provide...
Learning-based techniques, especially advanced pre-trained models for code have demonstrated capabilities in code understanding and generation, solving diverse software engineering (SE) tasks. Despite the promising results, current training approaches may not fully optimize model performance, as they typically involve learning from randomly shuffled training data. Recent work shows that Curriculum Learning (CL) can improve performance on code-related tasks through incremental learning based on the difficulty of synthetic code. Yet, the effectiveness of CL with conventional difficulty measures in SE tasks remains largely unexplored. In this study, we explore two conventional code metrics: code length and cyclomatic complexity to determine the difficulty levels. We investigate how the pre-trained code model (CodeT5) learns under CL, through the tasks of code clone detection and code summarization. Our empirical study on the CodeXGLUE benchmark showed contrasting results to prior studies, where the model exhibited signs of catastrophic forgetting and shortcut learning. Surprisingly, model performance saturates after only the first quartile of training, potentially indicating a limit in the model's representation capacity and/or the task's inherent difficulty. Future work should further explore various CL strategies with different code models across a wider range of SE tasks for a more holistic understanding.
Large Language Models (LLMs) are becoming increasingly competent across various domains, educators are showing a growing interest in integrating these LLMs into the learning process. Especially in software engineering, LLMs have demonstrated qualitatively better capabilities in code summarization, code generation, and debugging. Despite various research on LLMs for software engineering tasks in practice, limited research captures the benefits of LLMs for pedagogical advancements and their impact on the student learning process. To this extent, we analyze 126 undergraduate students' interaction with an AI assistant during a 13-week semester to understand the benefits of AI for software engineering learning. We analyze the conversations, code generated, code utilized, and the human intervention levels to integrate the code into the code base. Our findings suggest that students prefer ChatGPT over CoPilot. Our analysis also finds that ChatGPT generates responses with lower computational complexity compared to CoPilot. Furthermore, conversational-based interaction helps improve the quality of the code generated compared to auto-generated code. Early adoption of LLMs in software engineering is crucial to remain competitive in the rapidly developing landscape. Hence, the next generation of software engineers must acquire the necessary skills to interact with AI to improve productivity.
In large-scale software development, understanding the functionality and intent behind complex codebases is critical for effective development and maintenance. While code summarization has been widely studied, existing methods primarily focus on smaller code units, such as functions, and struggle with larger code artifacts like files and packages. Additionally, current summarization models tend to emphasize low-level implementation details, often overlooking the domain and business context that are crucial for real-world applications. This paper proposes a two-step hierarchical approach for repository-level code summarization, tailored to business applications. First, smaller code units such as functions and variables are identified using syntax analysis and summarized with local LLMs. These summaries are then aggregated to generate higher-level file and package summaries. To ensure the summaries are grounded in business context, we design custom prompts that capture the intended purpose of code artifacts based on the domain and problem context of the business application. We evaluate our approach on a business support system (BSS) for the telecommunications domain, showing that syntax analysis-based hierarchical summarization improves coverage, while business-context grounding enhances the relevance of the generated summaries.
Pre-training a language model and then fine-tuning it has shown to be an efficient and effective technique for a wide range of code intelligence tasks, such as code generation, code summarization, and vulnerability detection. However, pretraining language models on a large-scale code corpus is computationally expensive. Fortunately, many off-the-shelf Pre-trained Code Models (PCMs), such as CodeBERT, CodeT5, CodeGen, and Code Llama, have been released publicly. These models acquire general code understanding and generation capability during pretraining, which enhances their performance on downstream code intelligence tasks. With an increasing number of these public pre-trained models, selecting the most suitable one to reuse for a specific task is essential. In this paper, we systematically investigate the reusability of PCMs. We first explore three intuitive model selection methods that select by size, training data, or brute-force fine-tuning. Experimental results show that these straightforward techniques either perform poorly or suffer high costs. Motivated by these findings, we explore learning-based model selection strategies that utilize pre-trained models without altering their parameters. Specifically, we train proxy models to gauge the performance of pre-trained models, and measure the distribution deviation between a model's latent features and the task's labels, using their closeness as an indicator of model transferability. We conduct experiments on 100 widely-used opensource PCMs for code intelligence tasks, with sizes ranging from 42.5 million to 3 billion parameters. The results demonstrate that learning-based selection methods reduce selection time to 100 seconds, compared to 2,700 hours with brute-force fine-tuning, with less than 6% performance degradation across related tasks.
Code summarization aims to generate concise natural language descriptions for source code. Deep learning has been used more and more recently in software engineering, particularly for tasks like code creation and summarization. Specifically, it appears that the most current Large Language Models with coding perform well on these tasks. Large Language Models (LLMs) have significantly advanced the field of code summarization, providing sophisticated methods for generating concise and accurate summaries of source code. This study aims to perform a comparative analysis of several open-source LLMs, namely LLaMA-3, Phi-3, Mistral, and Gemma. These models' performance is assessed using important metrics such as BLEU\textsubscript{3.1} and ROUGE\textsubscript{3.2}. Through this analysis, we seek to identify the strengths and weaknesses of each model, offering insights into their applicability and effectiveness in code summarization tasks. Our findings contribute to the ongoing development and refinement of LLMs, supporting their integration into tools that enhance software development and maintenance processes.
Large language models have demonstrated promising performance across various software engineering tasks. While fine-tuning is a common practice to adapt these models for downstream tasks, it becomes challenging in resource-constrained environments due to increased memory requirements from growing trainable parameters in increasingly large language models. We introduce \approach, a technique to adapt large models for downstream code tasks using Code Property Graphs (CPGs). Our approach introduces a modular component called \transducer that enriches code embeddings with structural and dependency information from CPGs. The Transducer comprises two key components: Graph Vectorization Engine (GVE) and Attention-Based Fusion Layer (ABFL). GVE extracts CPGs from input source code and transforms them into graph feature vectors. ABFL then fuses those graphs feature vectors with initial code embeddings from a large language model. By optimizing these transducers for different downstream tasks, our approach enhances the models without the need to fine-tune them for specific tasks. We have evaluated \approach on three downstream tasks: code summarization, assert generation, and code translation. Our results demonstrate competitive performance compared to full parameter fine-tuning while reducing up to 99\% trainable parameters to save memory. \approach also remains competitive against other fine-tuning approaches (e.g., LoRA, Prompt-Tuning, Prefix-Tuning) while using only 1.5\%-80\% of their trainable parameters. Our findings show that integrating structural and dependency information through Transducer Tuning enables more efficient model adaptation, making it easier for users to adapt large models in resource-constrained settings.
Language models have proven successful across a wide range of software engineering tasks, but their significant computational costs often hinder their practical adoption. To address this challenge, researchers have begun applying various compression strategies to improve the efficiency of language models for code. These strategies aim to optimize inference latency and memory usage, though often at the cost of reduced model effectiveness. However, there is still a significant gap in understanding how these strategies influence the efficiency and effectiveness of language models for code. Here, we empirically investigate the impact of three well-known compression strategies -- knowledge distillation, quantization, and pruning -- across three different classes of software engineering tasks: vulnerability detection, code summarization, and code search. Our findings reveal that the impact of these strategies varies greatly depending on the task and the specific compression method employed. Practitioners and researchers can use these insights to make informed decisions when selecting the most appropriate compression strategy, balancing both efficiency and effectiveness based on their specific needs.