[TODO] Implementation of `Proto Splitter` (breakthrough of 2GB contract for Protocol Buffers)

[PINTO0309]

Issue Type

Others

OS

Linux

onnx2tf version number

1.15.12

onnx version number

1.13.1

onnxruntime version number

1.15.1

onnxsim (onnx_simplifier) version number

0.4.33

tensorflow version number

1.13.0

Download URL for ONNX

N/A

Parameter Replacement JSON

N/A

Description

1. Personal
2. Implementation of Proto Splitter (breakthrough of 2GB contract for Protocol Buffers) https://github.com/tensorflow/tensorflow/tree/master/tensorflow/tools/proto_splitter https://github.com/tensorflow/tensorflow/blob/master/tensorflow/tools/proto_splitter/g3doc/in-depth-guide.md https://github.com/tensorflow/community/blob/master/rfcs/20230720-unbound-saved-model.md

Implementing this in the backend should make it possible to break through the 2GB limit of Protocol Buffers.

TensorFlow v2.15.0+

[PINTO0309]

• TensorFlow v2.15.0, v2.16.1, v2.17.0

  ERROR: The option `experimental_image_format` is disabled in OSS.
  Traceback (most recent call last):
  File "/home/b920405/git/onnx2tf/onnx2tf/onnx2tf.py", line 1173, in convert
    tf.saved_model.save(model, output_folder_path, options=tf.saved_model.SaveOptions(experimental_image_format=True))
  File "/home/b920405/.local/lib/python3.10/site-packages/tensorflow/python/saved_model/save_options.py", line 189, in __init__
    raise ValueError(
  ValueError: The option `experimental_image_format` is disabled in OSS.

• test code https://github.com/tensorflow/tensorflow/blob/master/tensorflow/python/saved_model/proto_splitter_save_test.py