def __init__(self, params): self.params = params train_args = { 'num_agent_train_steps_per_iter': params['num_agent_train_steps_per_iter'], 'num_critic_updates_per_agent_update': params['num_critic_updates_per_agent_update'], 'train_batch_size': params['batch_size'], 'double_q': params['double_q'], 'device': params['device'], } env_args = get_env_kwargs(params['env_name']) self.agent_params = {**train_args, **env_args, **params} self.params['agent_class'] = DQNAgent self.params['agent_params'] = self.agent_params self.params['train_batch_size'] = params['batch_size'] self.params['env_wrappers'] = self.agent_params['env_wrappers'] self.rl_trainer = RL_Trainer(self.params)
def __init__(self, params): self.params = params train_args = { "num_agent_train_steps_per_iter": params["num_agent_train_steps_per_iter"], "num_critic_updates_per_agent_update": params[ "num_critic_updates_per_agent_update" ], "train_batch_size": params["batch_size"], "double_q": params["double_q"], } env_args = get_env_kwargs(params["env_name"]) self.agent_params = {**train_args, **env_args, **params} self.params["agent_class"] = ExplorationOrExploitationAgent self.params["agent_params"] = self.agent_params self.params["train_batch_size"] = params["batch_size"] self.params["env_wrappers"] = self.agent_params["env_wrappers"] self.rl_trainer = RL_Trainer(self.params)