Ramanujan Challenge paper reports an AI pipeline for proof judging
The pipeline cut false-positive verdicts on natural-language proofs from 23% to 6% versus one-shot LLM judgment.
Stories per day · 30 days
1 stories, newest first · one card per story