New RENEW framework uses human preferences to fix AI world models

Researchers have introduced RENEW, a novel framework designed to improve world models in offline reinforcement learning by using human preferences to correct exploitable dynamics. This method, termed Dynamics Learning from Human Feedback (DLHF), focuses on training the model to recognize and avoid "hallucinations" in its predictions. RENEW enhances sample efficiency by directing finetuning efforts towards areas where the model exhibits the most uncertainty and exploitable weaknesses, as demonstrated in experiments with Jumanji and classic control environments. AI

IMPACT This research offers a new approach to improving the robustness and reliability of AI world models in reinforcement learning scenarios.

RANK_REASON The cluster contains a research paper detailing a new method for AI model training. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• arXiv
• Bradley--Terry model
• Dynamics Learning from Human Feedback
• Jumanji
• RENEW

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →