Loading...
Market
Which AI research award will DeepMind's GenRM win by end of 2025?
2
Google DeepMind•University of Toronto•MILA•UCLA•Generative Reward Modeling•DeepMind•Large Language Models
Resolution / Starting Odds
AAAI • 25%
NeurIPS • 25%
ICML • 25%
Other • 25%
Official announcements from AI research award organizations
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
Turing Award • 25%
NeurIPS Best Paper Award • 25%
AAAI Innovative Application Award • 25%
Other • 25%
Best Game Design • 25%
Best Technology • 25%
Best Indie Game • 25%
Other • 25%
Yes • 50%
No • 50%
Turing Award • 25%
NeurIPS Competition • 25%
ICML Award • 25%
Other • 25%
Best AI Startup • 25%
Best Architectural Innovation • 25%
Best Use of AI in Design • 25%
Other • 25%
Google • 25%
OpenAI • 25%
DeepMind • 25%
Other • 25%
Grok 2 • 25%
Imagen 3 • 25%
GPT-5 • 25%
Other • 25%
Wins Best Innovation • 25%
Wins Best Performance • 25%
Wins Best Cost Efficiency • 25%
No Major Award • 25%