Noted
CWE-bench v2: More coverage, quality, and transparency
Updated benchmarks help practitioners evaluate AI models more accurately, though the provided excerpt lacks details on the benchmark's methodology or results.
Reported by Hugging Face ·
What happened
A blog post on Hugging Face announces CWE-bench v2, an updated benchmark aiming for more coverage, quality, and transparency. The release includes a public dataset and an associated leaderboard Space on the platform.
Who it’s about
- Hugging Face
- CWE-bench