Tag
Starting a blog post series on modern RL algorithms from the ground up, Part 1 covers the REINFORCE estimator, deriving unbiased policy gradients and analyzing variance.