Function epsilon_greedy

rl/q_learning.py:20–25

(Q, s, eps=0.1)

def epsilon_greedy(Q, s, eps=0.1):
  if np.random.random() < eps:
    return np.random.choice(ALL_POSSIBLE_ACTIONS)
  else:
    a_opt = max_dict(Q[s])[0]
    return a_opt


if __name__ == '__main__':