← Back to context

Comment by iwontberude

5 days ago

Yeah, unless the model is evaluated with blind hold outs, the benchmarks are misleading.

0 comments

iwontberude

Reply

No comments yet

Contribute on Hacker News ↗