Breakout room discussions
What do we think is difficult in RL
What types of tasks should we be solving?
Should we be using language models for RL?
Should we make an RL library for Mila?
Are there hurdles at Mila that are slowing down research?