9abf7a10295c16cecb76acb9472fadbfe4ca7c9a,safe_learning/tests/test_rl.py,TestPolicyIteration,test_integration,#TestPolicyIteration#,30

Before Change

        value_iter = rl.value_iteration()

        adapt_policy = tf.train.GradientDescentOptimizer(0.01).minimize(
            -tf.reduce_sum(rl.future_values(rl.state_space)),
            var_list=rl.policy.parameters)

After Change

            value_iter = rl.value_iteration()

            loss = -tf.reduce_sum(rl.future_values(rl.state_space))
            optimizer = tf.train.GradientDescentOptimizer(0.01)
            adapt_policy = optimizer.minimize(loss,
                                              var_list=rl.policy.parameters)

            sess.run(tf.global_variables_initializer())

Instances