NeuralBridge

Easy-to-use, one-liner AI model conversion tool between AI frameworks to ensure faster model development and portability with Open Neural Network Exchange (ONNX).

How to Install

To install NeuralBridge simple use:

pip install neuralbridge

In case you would like to build this project from source please consult the requirements.txt file and the src directory.

How to Use

The main use case of NeuralBridge is to convert model runtime. It converts TensorFlow and PyTorch models from one framework to another absolutely automatically. You can access runtime conversion like this:

from neuralbridge import convert_model

model_in_other_framework = convert_model(model_in_one_framework)

Examples

You can find additional examples in the examples directory.

Converting PyTorch model into TensorFlow

This example is in the example_tf_to_torch.py file.

Converting TensorFlow model into PyTorch

This example is in the example_torch_to_tf.py file.

Old design patterns

You can find the old design patterns in the following files:

• old_way_onnx_to_tf.py

• old_way_onnx_to_torch.py

• old_way_tf_to_onnx.py

• old_way_tf_to_torch.py

• old_way_torch_to_onnx.py

• old_way_torch_to_tf.py

Requirements

The list of all requirements are in the requirements.txt.

License

See details in the LICENSE.