def testSaveLessThanRestore(self):
try:
self._saveAndRestore(2, 3)
self.fail()
except tf.errors.NotFoundError:
pass
def testDynamicSaveRestore(self):
self._saveAndRestore(2, 2, saver_class=OptimisticRestoreSaver)
def testDynamicSaveMoreThanRestore(self):
self._saveAndRestore(3, 2, saver_class=OptimisticRestoreSaver)
def testDynamicSaveLessThanRestore(self):
self._saveAndRestore(2, 3, saver_class=OptimisticRestoreSaver)
def testDynamicSaveLessThanRestoreThenSaveThenLoadNormal(self):
self._saveAndRestore(2, 3, 3, saver_class=OptimisticRestoreSaver)
def testDynamicSaveLessThanRestoreThenSaveThenLoadMoreNormal(self):
try:
self._saveAndRestore(2, 3, 4, saver_class=OptimisticRestoreSaver)
self.fail()
except tf.errors.NotFoundError:
pass
if __name__ == '__main__':
# unittest.main()
tftest.main()
batch_size=3,
)
trajectories = [
self._make_trajectory(observations, actions) # pylint: disable=g-complex-comprehension
for (observations, actions) in [
(np.array([[0, 1]]), np.array([0])),
(np.array([[1, 2], [3, 4]]), np.array([0, 0])),
(np.array([[1, 2], [3, 4], [5, 6]]), np.array([0, 0, 0])),
]
]
metrics = simple.evaluate_model(env, trajectories, plt)
self.assertIsNotNone(metrics)
self.assertEqual(len(metrics), 2)
def test_fails_to_evaluate_model_with_matrix_observation_space(self):
with backend.use_backend('numpy'):
env = self._make_env( # pylint: disable=no-value-for-parameter
observation_space=gym.spaces.Box(shape=(2, 2), low=0, high=1),
action_space=gym.spaces.Discrete(n=1),
max_trajectory_length=2,
batch_size=1,
)
trajectories = [
self._make_trajectory(np.array([[0, 1], [2, 3]]), np.array([0]))]
metrics = simple.evaluate_model(env, trajectories, plt)
self.assertIsNone(metrics)
if __name__ == '__main__':
test.main()
= self._make_constant_displacement_image(
u, d, floating_data)
resampled_data = sess.run(
warped,
feed_dict={
img: floating_data\
.reshape(image_batch_shape),
disp: displacement_data\
.reshape(disp_batch_shape),
})
resampled_data = resampled_data.reshape(
floating_shape)
self._test_resampled(resampled_data, floating_data,
u, code, d)
def test_cpu_resampling(self):
self._test_resampling(False)
def test_gpu_resampling(self):
if tft.is_gpu_available(cuda_only=True) and tft.is_built_with_cuda():
self._test_resampling(True)
else:
self.skipTest('No CUDA support available')
if __name__ == '__main__':
tft.main()
