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

 (arxiv.org)

3 points | by zdw 7 hours ago ago

No comments yet.