Our Mission

Build safe, robust learning algorithms that can make use of experiential data efficiently to have a positive impact on the society

Three essential pilars of the lab

  • Efficient RL
  • Sample efficient learning algorithms
  • Model Recycling
  • Efficient sequence models
  • RLHF/Alignment
  • Uncertainty aware/Bayesian algorithms
  • Offline RL
  • Active learning/Human in the loop algorithms
  • Better evaluations
  • Improving reasoning
  • Creativity
  • Deliberation
  • Causality
  • Imagination
  • Planning