MineValiCoder introduces a closed-loop test-driven development framework for code generation from natural-language requirements. Its Test Case Quality Mining module self-validates generated tests and filters faulty supervision; Parallel TDD Refinement then iteratively improves code while producing diverse candidates; finally, Bipartite Graph-Based Code-Test Mutual Validation models code-test interactions to select code more stably. The abstract reports Pass@1 scores of 96.34% on HumanEval, 87.40% on MBPP, 64.00% on APPS, and 51.33% on LiveCodeBench across four LLMs, claiming improvements over state-of-the-art methods.
No heat snapshots are available in the last 24 hours.