search

openai / universe-starter-agent

A starter agent that can solve a number of universe environments.
MIT License
1.1k stars 318 forks source link

Easiest way to have an agent playing with the most likely action (no exploration / learning)? #106

Closed crci closed 6 years ago

crci commented 7 years ago

As I put in the title, I was wondering if there is a simple way of having one of the agents just playing as best as possible (i.e. taking always the most likely action), with no exploration (and even no learning). This would be useful to check the real skill level that the agent has achieved so far.

Thanks!

ethancaballero commented 7 years ago

have separate test thread that takes argmax (instead of multinomial sample) of policy.