Reshape#
Reshape - 14#
Version
name: Reshape (GitHub)
domain: main
since_version: 14
function: False
support_level: SupportType.COMMON
shape inference: True
This version of the operator has been available since version 14.
Summary
Reshape the input tensor similar to numpy.reshape. First input is the data tensor, second input is a shape tensor which specifies the output shape. It outputs the reshaped tensor. At most one dimension of the new shape can be -1. In this case, the value is inferred from the size of the tensor and the remaining dimensions. A dimension could also be 0, in which case the actual dimension value is unchanged (i.e. taken from the input tensor). If ‘allowzero’ is set, and the new shape includes 0, the dimension will be set explicitly to zero (i.e. not taken from input tensor). Shape (second input) could be an empty shape, which means converting to a scalar. The input tensor’s shape and the output tensor’s shape are required to have the same number of elements.
Attributes
allowzero: (Optional) By default, when any value in the ‘shape’ input is equal to zero the corresponding dimension value is copied from the input tensor dynamically. allowzero=1 indicates that if any value in the ‘shape’ input is set to zero, the zero value is honored, similar to NumPy.
Inputs
data (heterogeneous) - T: An input tensor.
shape (heterogeneous) - tensor(int64): Specified shape for output.
Outputs
reshaped (heterogeneous) - T: Reshaped data.
Type Constraints
T in ( tensor(bfloat16), tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.
Examples
_reshape
import numpy as np
import onnx
original_shape = [2, 3, 4]
test_cases = {
"reordered_all_dims": np.array([4, 2, 3], dtype=np.int64),
"reordered_last_dims": np.array([2, 4, 3], dtype=np.int64),
"reduced_dims": np.array([2, 12], dtype=np.int64),
"extended_dims": np.array([2, 3, 2, 2], dtype=np.int64),
"one_dim": np.array([24], dtype=np.int64),
"negative_dim": np.array([2, -1, 2], dtype=np.int64),
"negative_extended_dims": np.array([-1, 2, 3, 4], dtype=np.int64),
"zero_dim": np.array([2, 0, 4, 1], dtype=np.int64),
"zero_and_negative_dim": np.array([2, 0, 1, -1], dtype=np.int64),
}
data = np.random.random_sample(original_shape).astype(np.float32)
for test_name, shape in test_cases.items():
node = onnx.helper.make_node(
"Reshape",
inputs=["data", "shape"],
outputs=["reshaped"],
)
reshaped = reshape_reference_implementation(data, shape)
expect(
node,
inputs=[data, shape],
outputs=[reshaped],
name="test_reshape_" + test_name,
)
_allowzero
import numpy as np
import onnx
original_shape = [0, 3, 4]
test_cases = {
"allowzero_reordered": np.array([3, 4, 0], dtype=np.int64),
}
data = np.random.random_sample(original_shape).astype(np.float32)
for test_name, shape in test_cases.items():
node = onnx.helper.make_node(
"Reshape",
inputs=["data", "shape"],
outputs=["reshaped"],
allowzero=1, # if allowzero=1, final shape = (3, 4, 0)
# if allowzero=0, final shape = (3, 4, 4)
)
reshaped = reshape_reference_implementation(data, shape, allowzero=1)
expect(
node,
inputs=[data, shape],
outputs=[reshaped],
name="test_reshape_" + test_name,
)
Reshape - 13#
Version
name: Reshape (GitHub)
domain: main
since_version: 13
function: False
support_level: SupportType.COMMON
shape inference: True
This version of the operator has been available since version 13.
Summary
Reshape the input tensor similar to numpy.reshape. First input is the data tensor, second input is a shape tensor which specifies the output shape. It outputs the reshaped tensor. At most one dimension of the new shape can be -1. In this case, the value is inferred from the size of the tensor and the remaining dimensions. A dimension could also be 0, in which case the actual dimension value is unchanged (i.e. taken from the input tensor). Shape (second input) could be an empty shape, which means converting to a scalar. The input tensor’s shape and the output tensor’s shape are required to have the same number of elements.
Inputs
data (heterogeneous) - T: An input tensor.
shape (heterogeneous) - tensor(int64): Specified shape for output.
Outputs
reshaped (heterogeneous) - T: Reshaped data.
Type Constraints
T in ( tensor(bfloat16), tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.
Reshape - 5#
Version
name: Reshape (GitHub)
domain: main
since_version: 5
function: False
support_level: SupportType.COMMON
shape inference: True
This version of the operator has been available since version 5.
Summary
Reshape the input tensor similar to numpy.reshape. First input is the data tensor, second input is a shape tensor which specifies the output shape. It outputs the reshaped tensor. At most one dimension of the new shape can be -1. In this case, the value is inferred from the size of the tensor and the remaining dimensions. A dimension could also be 0, in which case the actual dimension value is unchanged (i.e. taken from the input tensor). Shape (second input) could be an empty shape, which means converting to a scalar. The input tensor’s shape and the output tensor’s shape are required to have the same number of elements.
Inputs
data (heterogeneous) - T: An input tensor.
shape (heterogeneous) - tensor(int64): Specified shape for output.
Outputs
reshaped (heterogeneous) - T: Reshaped data.
Type Constraints
T in ( tensor(bool), tensor(complex128), tensor(complex64), tensor(double), tensor(float), tensor(float16), tensor(int16), tensor(int32), tensor(int64), tensor(int8), tensor(string), tensor(uint16), tensor(uint32), tensor(uint64), tensor(uint8) ): Constrain input and output types to all tensor types.
Reshape - 1#
Version
name: Reshape (GitHub)
domain: main
since_version: 1
function: False
support_level: SupportType.COMMON
shape inference: False
This version of the operator has been available since version 1.
Summary
Reshape the input tensor similar to numpy.reshape. It takes a tensor as input and an argument shape. It outputs the reshaped tensor. At most one dimension of the new shape can be -1. In this case, the value is inferred from the size of the tensor and the remaining dimensions. A dimension could also be 0, in which case the actual dimension value is unchanged (i.e. taken from the input tensor). Shape (second input) could be an empty shape, which means converting to a scalar. The input tensor’s shape and the output tensor’s shape are required to have the same number of elements.
Attributes
consumed_inputs: legacy optimization attribute.
shape: New shape
Inputs
data (heterogeneous) - T: An input tensor.
Outputs
reshaped (heterogeneous) - T: Reshaped data.
Type Constraints
T in ( tensor(double), tensor(float), tensor(float16) ): Constrain input and output types to float tensors.