CodeAssistBench (CAB): Dataset & Benchmarking for Multi-turn Chat-Based Code Assistance

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kim, Myeongsoo, Garg, Shweta, Ray, Baishakhi, Kumar, Varun, Deoras, Anoop
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866915730513985536
author Kim, Myeongsoo
Garg, Shweta
Ray, Baishakhi
Kumar, Varun
Deoras, Anoop
author_facet Kim, Myeongsoo
Garg, Shweta
Ray, Baishakhi
Kumar, Varun
Deoras, Anoop
contents Programming assistants powered by large language models have improved dramatically, yet existing benchmarks still evaluate them in narrow code-generation settings. Recent efforts such as InfiBench and StackEval rely on Stack Overflow questions and remain limited to single-turn interactions, manually curated data, and isolated snippets rather than full project environments. We introduce CodeAssistBench (CAB), the first benchmark for evaluating multi-turn, project-grounded programming assistance at scale. CAB automatically constructs datasets from GitHub issues tagged as questions, using an LLM-driven pipeline that filters noise, extracts runnable contexts, builds executable containers, and verifies environment correctness. This enables continuous, automated expansion across diverse repositories without manual intervention. Using CAB, we create a testbed of 3,286 real-world issues across 214 repositories, spanning seven languages. Evaluating state-of-the-art models reveals a substantial gap: while models achieve 70-83% accuracy on Stack Overflow-style questions, they solve only 7.22-16.49% of CAB issues from post-training-cutoff repositories. These results highlight a fundamental challenge: current LLMs struggle to provide assistance in realistic, project-specific contexts despite strong performance on traditional Q&A benchmarks. CAB provides a scalable, reproducible framework for advancing research in multi-turn, codebase-grounded programming agents. The benchmark and pipeline are fully automated and publicly available at https://github.com/amazon-science/CodeAssistBench/.
format Preprint
id arxiv_https___arxiv_org_abs_2507_10646
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CodeAssistBench (CAB): Dataset & Benchmarking for Multi-turn Chat-Based Code Assistance
Kim, Myeongsoo
Garg, Shweta
Ray, Baishakhi
Kumar, Varun
Deoras, Anoop
Software Engineering
Artificial Intelligence
Programming assistants powered by large language models have improved dramatically, yet existing benchmarks still evaluate them in narrow code-generation settings. Recent efforts such as InfiBench and StackEval rely on Stack Overflow questions and remain limited to single-turn interactions, manually curated data, and isolated snippets rather than full project environments. We introduce CodeAssistBench (CAB), the first benchmark for evaluating multi-turn, project-grounded programming assistance at scale. CAB automatically constructs datasets from GitHub issues tagged as questions, using an LLM-driven pipeline that filters noise, extracts runnable contexts, builds executable containers, and verifies environment correctness. This enables continuous, automated expansion across diverse repositories without manual intervention. Using CAB, we create a testbed of 3,286 real-world issues across 214 repositories, spanning seven languages. Evaluating state-of-the-art models reveals a substantial gap: while models achieve 70-83% accuracy on Stack Overflow-style questions, they solve only 7.22-16.49% of CAB issues from post-training-cutoff repositories. These results highlight a fundamental challenge: current LLMs struggle to provide assistance in realistic, project-specific contexts despite strong performance on traditional Q&A benchmarks. CAB provides a scalable, reproducible framework for advancing research in multi-turn, codebase-grounded programming agents. The benchmark and pipeline are fully automated and publicly available at https://github.com/amazon-science/CodeAssistBench/.
title CodeAssistBench (CAB): Dataset & Benchmarking for Multi-turn Chat-Based Code Assistance
topic Software Engineering
Artificial Intelligence
url https://arxiv.org/abs/2507.10646