Planning to Explore: Curiosity-Driven Planning for LLM Test Generation

1. 1Identify the limitations of current greedy approaches to LLM-based test generation
2. 2Apply curiosity-driven planning to prioritize exploration of deep branches in the codebase
3. 3Use Bayesian principles to guide the planning process and maximize long-term coverage gain
4. 4Evaluate the effectiveness of the proposed approach in improving test coverage and reducing plateaus

This research benefits AI engineers and ML researchers working on LLM-based test generation, as it provides a novel approach to improving coverage of complex codebases

💡 Curiosity-driven planning can help overcome the limitations of greedy approaches to LLM-based test generation by prioritizing exploration of deep branches

🤖 Curiosity-driven planning for LLM-based test generation: a novel approach to improve coverage of complex codebases 💻