Number of episodes from the environment


How many episodes can an agent play within the 4 days training period?

I am asking because some RL methods need many episodes to improve the
policy. We try to calculate how many samples can be generated for
Q-learning (e.g. DQN), policy gradient, or SARSA.

Thanks for organizing this amazing challenge :smiley:

@GrizzlyRL2019: More details to be announced soon. But you will be allowed a total of 7 million samples during the training.