Co-evolving judge models bridge the policy model to human intent and boost policy performance with minimal human effort. Thanks @Zewei__Zhou for your amazing work during your internship! 💚
🤔 𝘾𝙖𝙣 𝙨𝙚𝙡𝙛-𝙞𝙢𝙥𝙧𝙤𝙫𝙚𝙢𝙚𝙣𝙩 𝙗𝙚 𝙨𝙪𝙨𝙩𝙖𝙞𝙣𝙚𝙙 𝙬𝙞𝙩𝙝 𝙖 𝙛𝙞𝙭𝙚𝙙, 𝙣𝙤𝙣-𝙘𝙤-𝙚𝙫𝙤𝙡𝙫𝙞𝙣𝙜 𝙟𝙪𝙙𝙜𝙚?
Better policies expose new failures. If their judges stay fixed, verification can become the bottleneck to further self-improvement.
00:00
