PolyTest introduces a novel approach that significantly improves automated unit test generation by exploiting polyglot capabilities and temperature-controlled diversity of large language models.

• Generates tests in multiple programming languages to enhance test quality
• Utilizes cross-prompt verification to ensure consistent, reliable test outputs
• Achieves higher code coverage and mutation scores than traditional single-attempt generation
• Demonstrates how strategic LLM prompting can overcome limitations in automated testing

This research matters for engineering teams seeking to automate quality assurance while reducing testing effort, potentially saving significant development time and improving software reliability.

Unify and Triumph: Polyglot, Diverse, and Self-Consistent Generation of Unit Tests with LLMs