onnx-source-snapshot/examples/np_array_tensorproto.ipynb

In [1]:

import numpy
import onnx
import os
from onnx import numpy_helper


# Preprocessing: create a Numpy array
numpy_array = numpy.array([[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]], dtype=float)
print("Original Numpy array:\n{}\n".format(numpy.array2string(numpy_array)))

Original Numpy array:
[[1. 2. 3.]
 [4. 5. 6.]]

In [2]:

# Convert the Numpy array to a TensorProto
tensor = numpy_helper.from_array(numpy_array)
print("TensorProto:\n{}".format(tensor))

TensorProto:
dims: 2
dims: 3
data_type: 11
raw_data: "\000\000\000\000\000\000\360?\000\000\000\000\000\000\000@\000\000\000\000\000\000\010@\000\000\000\000\000\000\020@\000\000\000\000\000\000\024@\000\000\000\000\000\000\030@"

In [3]:

# Convert the TensorProto to a Numpy array
new_array = numpy_helper.to_array(tensor)
print("After round trip, Numpy array:\n{}\n".format(numpy.array2string(numpy_array)))

After round trip, Numpy array:
[[1. 2. 3.]
 [4. 5. 6.]]

In [4]:

# Save the TensorProto
with open(os.path.join("resources", "tensor.pb"), "wb") as f:
    f.write(tensor.SerializeToString())

In [5]:

# Load the TensorProto
new_tensor = onnx.TensorProto()
with open(os.path.join("resources", "tensor.pb"), "rb") as f:
    new_tensor.ParseFromString(f.read())
print("After saving and loading, new TensorProto:\n{}".format(new_tensor))

After saving and loading, new TensorProto:
dims: 2
dims: 3
data_type: 11
raw_data: "\000\000\000\000\000\000\360?\000\000\000\000\000\000\000@\000\000\000\000\000\000\010@\000\000\000\000\000\000\020@\000\000\000\000\000\000\024@\000\000\000\000\000\000\030@"