3a0f8d18624b326de0400bff84ddc0f8c60c46e9,src/garage/tf/algos/reps.py,REPS,_policy_opt_input_values,#REPS#,519

Before Change

        // pylint: disable=unexpected-keyword-arg
        policy_opt_input_values = self._policy_opt_inputs._replace(
            obs_var=samples_data["observations"],
            action_var=samples_data["actions"],
            reward_var=samples_data["rewards"],
            valid_var=samples_data["valids"],
            feat_diff=self._feat_diff,

After Change

            list(np.ndarray): Flatten policy optimization input values.

        
        agent_infos = episodes.padded_agent_infos
        policy_state_info_list = [
            agent_infos[k] for k in self.policy.state_info_keys
        ]

        actions = [
            self._env_spec.action_space.flatten_n(act)
            for act in episodes.actions_list
        ]
        padded_actions = episodes.pad_to_last(np.concatenate(actions))

        // pylint: disable=unexpected-keyword-arg
        policy_opt_input_values = self._policy_opt_inputs._replace(
            obs_var=episodes.padded_observations,
            action_var=padded_actions,
            reward_var=episodes.padded_rewards,
            valid_var=episodes.valids,
            feat_diff=self._feat_diff,
            param_eta=self._param_eta,

Instances