John Schulman on creating TRPO

Off topic, but he resembles me physically to such a comical degree that my kids laughed about it when I showed them a video of him.

This episode was so good

I read the CPI paper thanks to the TRPO reference. For a 2002 paper, CPI approach was surprisingly useful to current RL paradigm once Schulman replaced the impractical policy mixing in the original

What’s it like doing this interview in public? Were there other customers watching? How many people were recording?

@grok is there a longer version of this episode?

Fine I’ll do it: @grok explain simply John Schulman’s TRPO paper

2013*

nice!

Damnn, your hair is so cool

Exciting to see how TRPO is evolving, Michael!

John was on the quarter zip swag before us 😭 is there a link to the full thing ?