Loading...
Market
Will DeepMind's GenRM be adopted by at least one major tech company other than Google by end of Q2 2025?
3
Google DeepMind•University of Toronto•MILA•UCLA•Generative Reward Modeling•DeepMind•Large Language Models
Resolution / Starting Odds
Yes • 50%
No • 50%
Official announcements from major tech companies such as Microsoft, Amazon, or Apple
Story
Google DeepMind's GenRM Enhances LLM Accuracy by Verifying Outputs
Sep 3, 2024, 01:04 PM
Researchers from Google DeepMind, University of Toronto, MILA, and UCLA have introduced a novel approach called Generative Reward Modeling (GenRM). DeepMind's GenRM improves the accuracy of Large Language Models (LLMs) by training them to verify their own outputs using next-token prediction and chain-of-thought (CoT) reasoning. The approach leverages the text generation capabilities of LLMs to improve their performance.
View original story
Similar markets
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Yes • 50%
No • 50%
Microsoft • 25%
Amazon • 25%
Meta • 25%
Other • 25%
