5f4fd31343ae302c0781ad50ac10208ecf09c0d0,tf_agents/bandits/policies/lin_ucb_policy.py,LinearUCBPolicy,__init__,#LinearUCBPolicy#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#Any#,49

Before Change

    self._dtype = self._data_vector[0].dtype
    self._expose_predicted_rewards = expose_predicted_rewards
    if expose_predicted_rewards:
      info_spec = PolicyInfo(
          predicted_rewards=tensor_spec.TensorSpec(
              [self._num_actions], dtype=self._dtype))
    else:
      info_spec = ()

    super(LinearUCBPolicy, self).__init__(
        time_step_spec=time_step_spec,
        action_spec=action_spec,
        info_spec=info_spec,

After Change

    self._dtype = self._data_vector[0].dtype
    self._emit_policy_info = emit_policy_info
    predicted_rewards = ()
    if policy_utilities.InfoFields.PREDICTED_REWARDS in emit_policy_info:
      predicted_rewards = tensor_spec.TensorSpec(
          [self._num_actions], dtype=self._dtype)
    info_spec = policy_utilities.PolicyInfo(predicted_rewards=predicted_rewards)

    super(LinearUCBPolicy, self).__init__(
        time_step_spec=time_step_spec,
        action_spec=action_spec,

Instances