ProgramBench: Can Language Models Rebuild Programs from Scratch?

github.com

3 pointsfittingopposite1 comment

I didn't managed to find the tests. How can we know that the tests are actually reasonable in this case ?