Over-optimization in RL is well-known, but it even occurs when KL(policy || base model) is constrained fairly tightly

Oct 11, 2024

Over-optimization in RL is well-known, but it even occurs when KL(policy || base model) is constrained fairly tightly

Posted by Cecile G. Tamura in category: policy

We’ve detected that JavaScript is disabled in this browser. Please enable JavaScript or switch to a supported browser to continue using x.com. You can see a list of supported browsers in our Help Center.

Help Center

0 comments

GETAS THREAT LEVEL: ELEVATED
FACEBOOK: 13,950 MEMBERS
LINKEDIN: 2,066 MEMBERS
TWITTER FEED: 31,494 MEMBERS
GETTR FEED: 39,482 MEMBERS

LIFEBOAT NEWS: 3,404 SUBSCRIBERS
GETAS ALERTS: 574 SUBSCRIBERS
BLOG: 122,211 POSTS
DONORS: 6,001

BOARDS: 2,940 MEMBERS
REPORTS: 74
PROGRAMS: 25
FORUMS: 24
QUOTES: 136

FIGHT AIDS: 3 MEMBERS
FOLDING@HOME: 15 MEMBERS
ROSETTA@HOME: 44 MEMBERS

Lifeboat Foundation

Safeguarding Humanity

Blog

Oct 11, 2024

Over-optimization in RL is well-known, but it even occurs when KL(policy || base model) is constrained fairly tightly

Posted by Cecile G. Tamura in category: policy

Leave a reply

Categories

Top 30 Authors

All Authors

Lifeboat Foundation

Safeguarding Humanity

Blog

Oct 11, 2024

Over-optimization in RL is well-known, but it even occurs when KL(policy || base model) is constrained fairly tightly

Posted by Cecile G. Tamura in category: policy

Leave a reply

Tag cloud

Categories

Top 30 Authors

All Authors

Blogroll