← Flash PapersWeak-to-Strong Generalization: Eliciting Strong Capabilities with Weak Supervision

Share this Flash Paper

Weak-to-Strong Generalization: Eliciting Strong Capabilities with Weak Supervision

https://flashpapers.ai/p/hrarug52he
Post to XPost to BlueskyShare on LinkedInSubmit to Hacker NewsSubmit to RedditEmail a link
Make your own
Collin Burns, Pavel Izmailov +10 more
cs.CL2023arXiv:2312.09390Code· 2.6k★· updated 2y ago
Details ▸
Performance Gap Recovered (PGR):nearly 80%· NLP BenchmarksPerformance Gap Recovered (PGR):roughly 10%· ChatGPT Reward Modeling