A compact software-engineering evaluation benchmark. It contains five realistic Python engineering tasks, each with a verified reference solution, and a small CLI that applies a candidate patch (human ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results