XREPOTEST: Benchmarking Multilingual Repository-Level Unit Test Generation for Large Language Models

Evaluation and Distillation of Source Code Generation Tasks by Large Language Models

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.

Optimizing Context and Cost in LLM ‐Based Unit Test Generation: A Study on External Dependency Retrieval Strategies

While Large Language Models (LLMs) have demonstrated remarkable capabilities in code generation, their effectiveness in unit testing is often constrained by insufficient context regarding external dependencies. This limitation is particularly pronounced in industrial settings, where proprietary code remains opaque to the model. To address this challenge, we present a systematic empirical study of multiple strategies for context enrichment and optimization in LLM‐based unit test generation, conducted on seven diverse projects (three open‐source and four proprietary industrial systems), encompassing 261 distinct methods. By evaluating seven implementations (ranging from basic prompts to optimized context reduction strategies) across 10 independent runs, we analysed a total of 28,710 test suites. Our results demonstrate that combining prompt engineering with external dependency retrieval achieves an average branch coverage increase of 11.52 percentage points on industrial software over the baseline, with statistically significant improvements across all competing implementations. Beyond coverage, richer context substantially reduces generation‐repair iterations, cutting median execution time by 51.3% in industrial projects. We further show that reducing external dependencies to method signatures alone decreases input token consumption by up to 46.6% (25.4% in industrial projects) while fully preserving the coverage and efficiency gains of the complete retrieval approach. To confirm that these benefits are not tied to a specific model, we replicate the core comparison across three LLM backends from different families, obtaining a consistent, statistically significant coverage improvement on industrial code in every case. These findings establish this optimized context strategy as a cost‐effective solution for scalable, industrial‐grade automated test generation.

Large Language Model Vulnerabilities

Capacity vs. architecture: an evaluation of SLMs for automated docstring generation

SWE-Bench ProMax: Benchmarking Agents on Large-Scale Multilingual Code Refactoring

As AI coding agents take on increasingly complex, long-horizon software engineering tasks, existing benchmarks are rapidly saturating and their evaluation quality has come under serious scrutiny: a recent audit found that nearly 60% of unsolved SWE-bench Verified instances contain flawed tests -- either overly narrow tests that reject correct solutions or overly broad tests that check unstated requirements -- and that frontier models can verbatim reproduce gold patches from training data. Code refactoring, which requires coordinated, behavior-preserving changes across many files, offers a substantially harder and more realistic test of agent capability, yet remains underserved by current benchmarks. We introduce SWE-Bench ProMax, an expert-curated, multilingual code refactoring benchmark of 170 instances drawn from real commits across seven programming languages (Python, Java, TypeScript, Go, C, C++, and Rust). Every instance undergoes rigorous, multi-stage curation that directly addresses the quality problems identified in prior benchmarks: issue descriptions are rewritten from scratch to provide precise, unambiguous specifications, and test suites are manually reviewed to remove overly narrow and overly broad tests. Tasks with insufficient complexity or limited cross-file scope are filtered out, yielding a benchmark of challenging, large-scale refactoring tasks that average 11.4 modified files and 261.6 lines of code per instance, substantially exceeding the scale of existing benchmarks. Experiments with frontier models under two agent scaffolds show that the best model achieves only 41.2% resolve rate, confirming that SWE-Bench ProMax presents a meaningful and unsaturated challenge for current AI coding agents. Our benchmark is available at https://huggingface.co/datasets/swe-bench-promax/SWE-Bench-ProMax.

Toward Reliable LLM Code Generation: Adaptive Routing Framework for Ambiguous Requirements

Natural-language requirements for program synthesis are often incomplete or ambiguous, yet large language models are commonly expected to generate code in a single pass. Prior clarification-based methods address this issue by asking follow-up questions when sampled candidate programs disagree, but fixed clarify-on-disagreement policies can overuse clarification and can also overtrust weak behavioral agreement. We present an adaptive routing framework for LLM-based program synthesis that treats clarification as an inference-time control decision. The framework augments a ClarifyGPT-style pipeline with execution-driven confidence estimation, semanticdifference analysis, and bounded candidate expansion, allowing the system to choose among direct generation, additional evidence gathering, and clarification. We evaluate the framework on MBPP, HumanEval, and extended-test variants using GPT-4.1 mini, Claude Haiku 4.5, and GPT-5.4 mini. Adaptive routing improves pass@1 accuracy by up to 7.60 percentage points over single-pass baselines. Compared with fixed-policy clarification, it preserves accuracy while reducing token usage by up to 57.2% for GPT-4.1 mini, and reallocates computation toward harder cases for Claude Haiku 4.5. These results suggest that clarification is most useful when triggered selectively based on execution evidence and semantic disagreement, even when ambiguity is observed indirectly through candidate behavior rather than through explicitly annotated ambiguous requirements.