SlopCodeBench: Benchmarking How Coding Agents Degrade over Long-Horizon Tasks

 (arxiv.org)

1 points | by FiberBundle 2 hours ago

1 comments