A small package to record execution graphs of neural networks in PyTorch.
The package uses hooks and the grad_fn attribute to record information.
This can be used to generate visualizations at different scope depths.
Licensed under MIT License. View documentation at https://torchrecorder.readthedocs.io/
Requirements:
- Python3.6+
- PyTorch v1.3 or greater (the
cpuversion) - The Graphviz library and
graphvizpython package.
Install this package:
$ pip install torchrecorder
This is inspired from szagoruyko/pytorchviz. This package
differs from pytorchviz as it provides rendering at multiple depths.
Note that for rendering a network during training, you can use TensorBoard and
torch.utils.tensorboard.SummaryWriter.add_graph,
which records and renders to a protobuf in a single step. The intended usage of torchrecorder is for
presentation purposes.