Combined views
883
2 posts, first seen 2h ago
10 likes saves
883
2 posts, first seen 2h ago
Ask a question below.
Published answers will appear here.
The company urges researchers to stop relying on the evaluation.
i challenge openai to release a new, high-quality coding eval. they clearly have good ones internally, as do their major competitors. i published three evals when i worked there so i know it’s possible! it’s easy to find flaws in evals, but it’s hard to build good ones https://twitter.com/OpenAI/status/2074972179385720836
Ask a question below.
Published answers will appear here.