SlopCodeBench: Benchmarking How Coding Agents Degrade over Long-Horizon Tasks(arxiv.org)
1 pointby FiberBundle9 hours ago1 comment