Hyperparameters

Algorithm	Hyperparameter	Value
Q-learning	temporal discount factor γ	0.9
Q-learning	TD(λ) decay factor γ	0.5
Q-learning	learning rate, α	0.1
Q-learning	neg. reward per step	0.01
SR	temporal discount factor γ	0.9
SR	TD(λ) decay factor γ	0.5
SR	learning rate, α	0.1
SARSA	temporal discount factor γ	0.99
SARSA	TD(λ) decay factor γ	0.5
SARSA	learning rate, α	0.1
SARSA	neg. reward per step	0.001
Tile coding	tile size	[2 × 2, 3 × 3]
MB-G	model buffer window, N	15