4427e66372ade50d4bc36a7cfe93bbdbf71234cc,ml/rl/test/gridworld/test_gridworld_continuous.py,TestGridworldContinuous,test_trainer_sarsa,#TestGridworldContinuous#,53

Before Change

    def test_trainer_sarsa(self):
        environment = GridworldContinuous()
        states, actions, propensities, rewards, next_states, next_actions, is_terminal, possible_next_actions, reward_timelines = environment.generate_samples(
            100000, 1.0
        )
        trainer = self.get_sarsa_trainer(environment)
        predictor = trainer.predictor()
        evaluator = GridworldContinuousEvaluator(environment, False)
        tdps = environment.preprocess_samples(
            states,
            actions,
            propensities,
            rewards,
            next_states,
            next_actions,
            is_terminal,
            possible_next_actions,
            reward_timelines,
            self.minibatch_size,
        )

        self.assertGreater(evaluator.evaluate(predictor), 0.15)

        for tdp in tdps:

After Change

        trainer = self.get_sarsa_trainer(environment)
        predictor = trainer.predictor()
        evaluator = GridworldContinuousEvaluator(environment, False)
        tdps = environment.preprocess_samples(samples, self.minibatch_size)

        self.assertGreater(evaluator.evaluate(predictor), 0.15)

        for tdp in tdps:

Instances