In a single-prompt study on one Ruby class, DeepSeek-Chat produced the most maintainable RSpec test skeletons, while GPT-4's high method coverage was undercut by incorrect RSpec conventions.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.SE 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Comparative Evaluation of Large Language Models for Test-Skeleton Generation
In a single-prompt study on one Ruby class, DeepSeek-Chat produced the most maintainable RSpec test skeletons, while GPT-4's high method coverage was undercut by incorrect RSpec conventions.