The UCL Deciding, Acting, and Reasoning with Knowledge (DARK) Lab is a Reinforcement Learning research group at the UCL Centre for Artificial Intelligence. We focus on research in complex open-ended environments that provide a constant stream of novel observations without reliable reward functions, often requiring agents to create their own curricula and to deal with external knowledge, natural language, and hard exploration problems.
News
-
25/09/2024:
Rainbow Teaming: Open-Ended Generation of Diverse Adversarial Prompts has been accepted to NeurIPS 2024.
-
25/09/2024:
JaxMARL: Multi-Agent RL Environments and Algorithms in JAX has been accepted to NeurIPS 2024.
-
23/07/2024:
We have been awarded two Best Paper Awards at ICML 2024 for Genie: Generative Interactive Environments and Debating with More Persuasive LLMs Leads to More Truthful Answers!
-
10/07/20:
8 papers accepted at ICML 2024, including 3 Orals! Check out here for more details.
-
19/02/2024:
IQL-TD-MPC: Implicit Q-Learning for Hierarchical Model Predictive Control has been accepted to ICRA 2024 (oral).
-
19/02/2024:
H-GAP: Humanoid Control with a Generalist Planner has been accepted to ICLR 2024 (spotlight).
-
19/02/2024:
Understanding the Effects of RLHF on LLM Generalisation and Diversity, Mechanistically analyzing the effects of fine-tuning on procedurally defined tasks and Reward Model Ensembles Help Mitigate Overoptimization have been accepted to ICLR 2024.
-
18/02/2024:
Multi-Agent Diagnostics for Robustness via Illuminated Diversity has been accepted to AAMAS 2024 (oral).