Overall, this case study demonstrates the feasibility of bringing strong LLM-based code completion to low-resource programming languages, with models small enough to provide ``real-time''in-IDE support.
Abstract
Large Language Models (LLMs) unlocked new possibilities in automated code writing, becoming the backbone of most code completion tools. While LLMs excel in mainstream languages, they often lack support for the so-called low-resource languages where training data is scarce. As a result, these languages lag behind in the quality of code completion tooling available to their communities. A concrete example is Pharo, a Smalltalk-inspired language whose IDE currently offers only single-token completion. In this work, we report on our experience bringing LLM-based code completion to Pharo. First, we describe an end-to-end pipeline that combines Pharo-specific data curation, continued pre-training and fine-tuning of open code LLMs. Second, we introduce a set of Pharo code completion benchmarks designed to evaluate whether models (i) learn Pharo's syntax and (ii) accurately complete masked Pharo code from real-world GitHub repositories. Third, we show empirically that Pharo-specialized models substantially outperform their original base checkpoints and also exceed the accuracy of substantially larger code LLMs on Pharo completion. Overall, our case study demonstrates the feasibility of bringing strong LLM-based code completion to low-resource programming languages, with models small enough to provide ``real-time''in-IDE support.
It is observed that generated code often omits basic input validation or memory-safety checks, which can lead to overflows, resource exhaustion, or other reliability/security issues, and even the largest models frequently make simple mistakes.
Rodrigo Pato Nogueira, Marco Vieira, João R. Campos· 0 citations
The findings show that, while LLMs achieve promising results, they struggle with harder problems and with programming languages that have fewer available resources for training, and they often fail due to fundamental and easily avoidable errors that underscore the unreliability of automatically generated code.
Rodrigo Pato Nogueira, Marco Vieira, João R. Campos· Empirical Software Engineeri...· 1 citation
Large Language Models (LLMs) are fundamentally transforming computer system research and development. As we employ LLMs in file system (fs) development, it is essential to understand their capabilities, limitations, and operational efficiency for domain-specific tasks. We present \phi-Bench, an LLM benchmarking framework for fs-specific tasks. To facilitate benchmarking, we develop six types of tasks in \phi-Bench: basic understanding, basic implementation, performance modeling, debugging, optimization, and new feature development. Each type emphasizes different LLM capabilities: instruction following, knowledge recall, reasoning, or coding. To create high-quality tasks while achieving broad coverage with minimal human effort, we develop a new AI-assisted task generation pipeline in addition to expert-written and textbook-adapted tasks. With 505 tasks in \phi-Bench, we conduct an empirical study with both open source (DeepSeek-V4-Flash, GLM-5.1, and MiniMax-M2.7) and proprietary (Claude-Opus-4.7, GPT-5.2, and Gemini-3.1-Pro) LLMs. Our study discloses the model efficiency for different tasks, causes of failed fs tasks, and techniques for mitigating LLM failures. We will open source \phi-Bench to facilitate public research on using LLMs for fs development.
Large language models (LLMs) excel at general-purpose code generation, yet how well they handle scientific code remains an open question. Existing datasets and benchmarks are limited in scale, domain coverage, or executable verification, leaving the true gap between current LLMs and reliable scientific code generators inadequately assessed. To address these limitations, we present SciCodePile, the largest scientific code corpus to date, constructed from 37,737 public repositories and collectively comprising 128GB of code that spans multiple computational science disciplines. From this corpus, we further curate an executable benchmark of 200 tasks, each equipped with a sandboxed execution environment and an automated test harness for functional verification. We evaluate 15 LLMs from both open-source and closed-source families on three tasks: prefix-to-suffix completion, fill-in-the-middle infilling, and executable code generation. Results show that scientific code generation remains highly challenging: The best CodeBLEU reaches only 38.13 and 38.37 on the two completion tasks, while the strongest model achieves just 12.30\% Pass@1 on the executable benchmark, underscoring how far current models remain from reliable scientific code generation. To demonstrate the training utility of SciCodePile, we further show that continued pretraining on our corpus improves CodeBLEU by $\times$2.84 on scientific code completion, and instruction tuning on our data improves Pass@1 by $\times$4.79 on the executable benchmark. All code and data are available at https://huggingface.co/SciCodePile.
Weifeng Sun, Ye Fan, Yuchen Chen et al.· 0 citations
Large Language Models (LLMs) are predominantly assessed based on their common sense reasoning, language comprehension, and logical reasoning abilities. While models trained in specialized domains like mathematics or coding have demonstrated remarkable advancements in logical reasoning, there remains a significant gap in evaluating their code generation capabilities. Existing benchmark datasets fall short in pinpointing specific strengths and weaknesses, impeding targeted enhancements in models’ reasoning abilities to synthesize code.
To bridge this gap, this thesis introduces two novel contributions: CodeEval and CodeQual. CodeEval is an innovative, pedagogical benchmarking method that mirrors the evaluation processes encountered in academic programming courses. It comprises a multi-dimensional benchmark dataset of 602 hand-crafted problems designed to rigorously evaluate LLMs across 24 distinct aspects of Python programming, covering three proficiency levels—beginner, intermediate, and advanced—and includes both class-based and function-based problem types with detailed problem specifications and comprehensive test suites achieving 99.1% coverage. To facilitate widespread adoption, we developed RunCodeEval, an open-source execution framework that provides researchers with a ready-to-use evaluation pipeline. Our evaluation of 15 state-of-the-art LLMs revealed consistent performance degradation with increasing complexity (validated statistically, Cohen’s d = 0.790) and universal struggles with advanced concepts like concurrency.
Code quality is inherently subjective, encompassing dimensions like readability, efficiency, and adherence to language idioms that traditional static metrics fail to capture adequately. While large language models can assess these subjective qualities, lightweight models offer practical advantages: seamless CI/CD pipeline integration, lower operational costs, and full control over model behavior. We investigate whether such models can learn to assess code quality by training on synthetic LLM annotations. We introduce CodeQual, a dataset of 5,819 code samples derived from five established sources spanning diverse domains—competitive programming, pedagogical problems, software engineering, and general benchmarks—scored by LLMs across five quality dimensions, with 655 human-annotated samples for evaluation. Our fine-tuned model, CodeQualBERT, not only matches LLM performance but exceeds inter-human agreement on all five dimensions, achieving 16–100% improvement over the inter-human agreement baseline.
Together, these contributions provide a comprehensive framework for evaluating and improving LLMs in software engineering contexts, encompassing both functional correctness assessment and subjective code quality evaluation.
Evaluating SEDCoT on a public COBOL-to-C dataset demonstrates that it outperforms state-of-the-art baselines by at least 12% while producing translations with substantially higher readability than rule-based alternatives.
Phillip Entin, Wenchao Gu, Alexander Knapp et al.· 0 citations