Python train_evalの例

コード例 #1

ファイル: train_eval_test.py プロジェクト: egonina/agents

 def testDQNCartPole(self):
     root_dir = self.get_temp_dir()
     train_loss = train_eval.train_eval(root_dir,
                                        num_iterations=1,
                                        num_eval_episodes=1,
                                        initial_collect_steps=10)
     self.assertGreater(train_loss.loss, 0.0)

コード例 #2

ファイル: train_eval_test.py プロジェクト: egonina/agents

 def testRNNDQNMaskedCartPole(self):
     root_dir = self.get_temp_dir()
     train_loss = train_eval.train_eval(root_dir,
                                        env_name='MaskedCartPole-v0',
                                        train_sequence_length=2,
                                        initial_collect_steps=10,
                                        num_eval_episodes=1,
                                        num_iterations=1)
     self.assertGreater(train_loss.loss, 0.0)

コード例 #3

  def testDQNCartPole(self):
    if not tf.executing_eagerly():
      self.skipTest('Binary is eager-only.')

    root_dir = self.get_temp_dir()
    train_loss = train_eval.train_eval(root_dir,
                                       num_iterations=1,
                                       num_eval_episodes=1,
                                       initial_collect_steps=10)
    self.assertGreater(train_loss.loss, 0.0)

コード例 #4

ファイル: train_eval_test.py プロジェクト: whghdrms/agents-tutorial

    def testRNNDQNMaskedCartPole(self):
        if not tf.executing_eagerly():
            self.skipTest('Binary is eager-only.')

        root_dir = self.get_temp_dir()
        train_loss = train_eval.train_eval(root_dir,
                                           env_name='MaskedCartPole-v0',
                                           train_sequence_length=2,
                                           initial_collect_steps=10,
                                           num_eval_episodes=1,
                                           num_iterations=1)
        self.assertGreater(train_loss.loss, 0.0)