To use Model Explorer, users need to install it using pip and then run the model-explorer command. The article also mentions a Wiki that provides more details about installation, user guide, command line guide, API guide, running in Colab Notebook, developing adapter extension, and known limitations and issues. However, the article states that the Model Explorer team is not accepting contributions at this time and will be the only ones contributing to the repository.
Key takeaways:
- Model Explorer provides a hierarchical visualization of model graphs, with features such as highlighting input and output operations, overlaying metadata on nodes, and GPU-accelerated graph rendering.
- It supports TFLite, TF, TFJS, MLIR, and PyTorch (Exported Program) model format, and also provides an extension framework for developers to add support for additional formats.
- To use Model Explorer, users need to install it using pip and can refer to the Wiki for guides on installation, usage, API, and more.
- Currently, the Model Explorer team is not accepting contributions and will be the ones to contribute to the repository.