def _observation_spec():
"""Returns the observation spec."""
return {
1:
dm_env_rpc_pb2.TensorSpec(name=_OBSERVATION_BOARD,
shape=[_NUM_ROWS, _NUM_COLUMNS],
dtype=dm_env_rpc_pb2.FLOAT),
2:
dm_env_rpc_pb2.TensorSpec(name=_OBSERVATION_REWARD,
dtype=dm_env_rpc_pb2.FLOAT)
}
def setUp(self):
super(SpecManagerTests, self).setUp()
specs = {
54:
dm_env_rpc_pb2.TensorSpec(name='fuzz',
shape=[2],
dtype=dm_env_rpc_pb2.DataType.FLOAT),
55:
dm_env_rpc_pb2.TensorSpec(name='foo',
shape=[3],
dtype=dm_env_rpc_pb2.DataType.INT32),
}
self._spec_manager = spec_manager.SpecManager(specs)
def test_duplicate_names_raise_error(self):
specs = {
54:
dm_env_rpc_pb2.TensorSpec(name='fuzz',
shape=[3],
dtype=dm_env_rpc_pb2.DataType.FLOAT),
55:
dm_env_rpc_pb2.TensorSpec(name='fuzz',
shape=[2],
dtype=dm_env_rpc_pb2.DataType.FLOAT),
}
with self.assertRaisesRegex(ValueError, 'duplicate name'):
spec_manager.SpecManager(specs)
def test_setting_spec(self): tensor_spec = dm_env_rpc_pb2.TensorSpec() tensor_spec.name = 'Foo' tensor_spec.min.floats.array[:] = [0.0] tensor_spec.max.floats.array[:] = [0.0] tensor_spec.shape[:] = [2, 2] tensor_spec.dtype = dm_env_rpc_pb2.DataType.FLOAT
def test_set_multiple_bounds(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', shape=(2,), dtype=dm_env_rpc_pb2.DataType.INT32)
tensor_spec_utils.set_bounds(tensor_spec, minimum=[1, 2], maximum=[3, 4])
self.assertEqual([1, 2], tensor_spec.min.int32s.array)
self.assertEqual([3, 4], tensor_spec.max.int32s.array)
def test_set_scalar_bounds_int8(self): tensor_spec = dm_env_rpc_pb2.TensorSpec(dtype=dm_env_rpc_pb2.DataType.INT8) minimum = 1 maximum = 2 tensor_spec_utils.set_bounds(tensor_spec, minimum=minimum, maximum=maximum) self.assertEqual(np.int8(minimum).tobytes(), tensor_spec.min.int8s.array) self.assertEqual(np.int8(maximum).tobytes(), tensor_spec.max.int8s.array)
def test_min_and_max_legacy(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.INT32
tensor_spec.min.int32 = -1
tensor_spec.max.int32 = 1
bounds = tensor_spec_utils.bounds(tensor_spec)
self.assertEqual((-1, 1), bounds)
def _action_spec():
"""Returns the action spec."""
return {
1:
dm_env_rpc_pb2.TensorSpec(dtype=dm_env_rpc_pb2.INT8,
name=_ACTION_PADDLE)
}
def test_min_and_max(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.INT32
tensor_spec.min.int32s.array[:] = [-1]
tensor_spec.max.int32s.array[:] = [1]
bounds = tensor_spec_utils.bounds(tensor_spec)
self.assertEqual((-1, 1), bounds)
def test_set_scalar_bounds(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', dtype=dm_env_rpc_pb2.DataType.INT32)
tensor_spec_utils.set_bounds(tensor_spec, minimum=1, maximum=2)
self.assertEqual([1], tensor_spec.min.int32s.array)
self.assertEqual([2], tensor_spec.max.int32s.array)
def test_invalid_min_shape(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.min.uint32s.array[:] = [1, 2]
with self.assertRaisesRegex(
ValueError, 'Scalar tensors must have exactly 1 element.*'):
tensor_spec_utils.bounds(tensor_spec)
def test_cannot_set_any_min_bounds_to_exceed_maximum(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', shape=[2], dtype=dm_env_rpc_pb2.DataType.INT32)
with self.assertRaisesRegex(ValueError, 'larger than max'):
tensor_spec_utils.set_bounds(tensor_spec,
minimum=[0, 4],
maximum=[1, 1])
def test_set_only_max(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', dtype=dm_env_rpc_pb2.DataType.INT32)
tensor_spec_utils.set_bounds(tensor_spec, minimum=None, maximum=1)
self.assertIsNone(tensor_spec.min.WhichOneof('payload'))
self.assertEqual([1], tensor_spec.max.int32s.array)
def test_cannot_set_multiple_max_bounds_on_variable_shape(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', shape=[2, -1], dtype=dm_env_rpc_pb2.DataType.INT32)
with self.assertRaisesRegex(ValueError, 'incompatible'):
tensor_spec_utils.set_bounds(tensor_spec,
minimum=1,
maximum=[2, 3])
def test_cannot_set_dissimilar_shape_on_max_bounds(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', shape=[2, 2], dtype=dm_env_rpc_pb2.DataType.INT32)
with self.assertRaisesRegex(ValueError, 'incompatible'):
tensor_spec_utils.set_bounds(tensor_spec,
minimum=0,
maximum=[1, 2, 3])
def test_cannot_set_nonnumeric_bounds(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', dtype=dm_env_rpc_pb2.DataType.STRING)
with self.assertRaisesRegex(ValueError, 'non-numeric'):
tensor_spec_utils.set_bounds(tensor_spec,
minimum=None,
maximum=None)
def test_can_set_broadcastable_max_bounds(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', shape=[2], dtype=dm_env_rpc_pb2.DataType.INT32)
tensor_spec_utils.set_bounds(tensor_spec, minimum=[2, 3], maximum=4)
self.assertEqual([2, 3], tensor_spec.min.int32s.array)
self.assertEqual([4], tensor_spec.max.int32s.array)
def test_infinite_bounds_are_valid_for_floats(self, minimum, maximum):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.DOUBLE
tensor_spec.min.double = minimum
tensor_spec.max.double = maximum
tensor_spec.name = 'foo'
tensor_spec_utils.bounds(tensor_spec)
def test_nonnumeric_type_raises_error(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.STRING
tensor_spec.max.int32s.array[:] = [1]
tensor_spec.name = 'foo'
with self.assertRaisesRegex(ValueError, 'foo.*non-numeric.*string'):
tensor_spec_utils.bounds(tensor_spec)
def test_max_mismatches_type_raises_error(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.max.int32s.array[:] = [1]
tensor_spec.name = 'foo'
with self.assertRaisesRegex(ValueError, 'foo.*uint32.*max.*int32'):
tensor_spec_utils.bounds(tensor_spec)
def test_no_bounds_gives_arrayspec(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.shape[:] = [3]
tensor_spec.name = 'foo'
actual = dm_env_utils.tensor_spec_to_dm_env_spec(tensor_spec)
self.assertEqual(specs.Array(shape=[3], dtype=np.uint32), actual)
self.assertEqual('foo', actual.name)
def test_string_give_string_array(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.STRING
tensor_spec.shape[:] = [1, 2, 3]
tensor_spec.name = 'string_spec'
actual = dm_env_utils.tensor_spec_to_dm_env_spec(tensor_spec)
self.assertEqual(specs.StringArray(shape=[1, 2, 3]), actual)
self.assertEqual('string_spec', actual.name)
def test_invalid_max_shape(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.max.uint32s.array[:] = [1, 2]
tensor_spec.shape[:] = (2, 2)
with self.assertRaisesRegex(
ValueError, 'cannot reshape array of size .* into shape.*'):
tensor_spec_utils.bounds(tensor_spec)
def test_max_broadcast(self): tensor_spec = dm_env_rpc_pb2.TensorSpec() tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32 tensor_spec.max.uint32s.array[:] = [1] tensor_spec.shape[:] = (2, 2) bounds = tensor_spec_utils.bounds(tensor_spec) np.testing.assert_array_equal(np.full(tensor_spec.shape, 0), bounds.min) np.testing.assert_array_equal(np.full(tensor_spec.shape, 1), bounds.max)
def test_max_less_than_min_raises_error(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.INT32
tensor_spec.max.int32s.array[:] = [-1]
tensor_spec.min.int32s.array[:] = [1]
tensor_spec.name = 'foo'
with self.assertRaisesRegex(ValueError, 'foo.*min 1.*max -1'):
tensor_spec_utils.bounds(tensor_spec)
def test_broadcast_var_shape(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.INT32
tensor_spec.min.int32s.array[:] = [-1]
tensor_spec.max.int32s.array[:] = [1]
tensor_spec.shape[:] = (-1, )
bounds = tensor_spec_utils.bounds(tensor_spec)
self.assertEqual((-1, 1), bounds)
def _action_spec():
"""Returns the action spec."""
paddle_action_spec = dm_env_rpc_pb2.TensorSpec(dtype=dm_env_rpc_pb2.INT8,
name=_ACTION_PADDLE)
tensor_spec_utils.set_bounds(paddle_action_spec,
minimum=np.min(_VALID_ACTIONS),
maximum=np.max(_VALID_ACTIONS))
return {1: paddle_action_spec}
def test_unset_min_and_max(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec(
name='test', dtype=dm_env_rpc_pb2.DataType.INT32)
tensor_spec_utils.set_bounds(tensor_spec, minimum=1, maximum=2)
tensor_spec_utils.set_bounds(tensor_spec, minimum=None, maximum=None)
self.assertIsNone(tensor_spec.min.WhichOneof('payload'))
self.assertIsNone(tensor_spec.max.WhichOneof('payload'))
def test_max_scalar_doesnt_broadcast(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.max.uint32s.array[:] = [1]
tensor_spec.shape[:] = (2, 2)
bounds = tensor_spec_utils.bounds(tensor_spec)
self.assertEqual(0, bounds.min)
self.assertEqual(1, bounds.max)
def test_bounds_wrong_type_gives_error(self):
tensor_spec = dm_env_rpc_pb2.TensorSpec()
tensor_spec.dtype = dm_env_rpc_pb2.DataType.UINT32
tensor_spec.shape[:] = [3]
tensor_spec.name = 'foo'
tensor_spec.min.floats.array[:] = [1.9]
with self.assertRaisesRegex(ValueError, 'uint32'):
dm_env_utils.tensor_spec_to_dm_env_spec(tensor_spec)
