SEAIPLNov 23, 2024

A Preliminary Study of Multilingual Code Language Models for Code Generation Task Using Translated Benchmarks

arXiv:2411.15470v11 citationsh-index: 7Has Code2024 39th IEEE/ACM International Conference on Automated Software Engineering Workshops (ASEW)
Originality Synthesis-oriented
AI Analysis

This addresses the problem of benchmark reliability for multilingual code generation, but it is incremental as it builds on existing translated benchmarks.

The study evaluated Poly-Coder, a multilingual code language model, on translated benchmarks for code generation, finding that results align with training metrics like perplexity but revealing inconsistencies and replication challenges.

Evaluating the performance of Code Language Models (CLMs) for software engineering tasks, especially in multilingual and low-resource programming language settings, poses significant challenges. These challenges are primarily due to the lack of high-quality benchmarks across various programming languages and the imbalanced nature of the CLMs training corpus. Although recent advances in one of the common downstream tasks, code generation, have shown promise by introducing translated benchmarks using different methodologies, there is a current lack of empirical evidence assessing these benchmarks. To address this gap, we conducted a preliminary study to evaluate the performance of Poly-Coder, a pioneering open-source, multilingual CLM built for code generation. We utilized two existing state-of-the-art translations of the popular code generation benchmark, HumanEval, facilitated by the OctoPack and MultiPL-E studies. Our results suggest that the outcomes observed in these translated benchmarks align well with evaluation metrics used during the training phase, such as perplexity, thereby validating their effectiveness in estimating the performance of CLMs. However, we identified several inconsistencies in the CLMs' performance across the translated benchmarks and encountered challenges in replicating the results. These initial insights highlight the need for more comprehensive empirical studies to fully understand translated benchmarks' methodological approaches, limitations, and reproducibility. Such studies are essential to ensure their reliability before they are widely adopted.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes