d3rlpy.online.explorers.ConstantEpsilonGreedy

class d3rlpy.online.explorers.ConstantEpsilonGreedy(epsilon)[source]

\(\epsilon\)-greedy explorer with constant \(\epsilon\).

Parameters

epsilon (float) – the constant \(\epsilon\).

Methods

sample(algo, x, step)[source]
Parameters
  • algo (d3rlpy.online.explorers._ActionProtocol) –

  • x (numpy.ndarray) –

  • step (int) –

Return type

numpy.ndarray