jagged stream
Hello, I have a question on ML agents, as I am starting out i want to know if certain things are possible.
I want to train an AI live with the player (everytime the AI dies it learns to respond to player actions) but im afraid it takes too long, so I decided to simulate a player and have that bot train my AI over a longer period of time.
Question part: I was planning to save policies of the training every few episodes and use those as a "rank-up" system where everytime the AI is reinstated it will progress to the next "better trained" policy.
Actual Question: Does Unity ML-Agents allow me to save the training values every x episodes?
full forum