The latest release of google-ai-edge/model-explorer is model-explorer-v0.1.33. Published at github.com/google-ai-edge/model-explorer, this project is a JavaScript tool that hierarchically visualizes a model's operation graph so you can read its structure, and it is distributed under the Apache-2.0 license. The repository was created in January 2024 and currently has 1581 stars.
Model Explorer organizes model operations into nested layers, so layers can be dynamically expanded or collapsed. It highlights input and output operations, displays node metadata, shows operation details in interactive popups, and also includes search and identical-layer display features. Graph rendering supports GPU acceleration. It can read the TFLite, TF, TFJS, MLIR and PyTorch (Exported Program) formats, and it also provides an extension framework for adding further formats. Analyzing the structure of multiple model formats in a single tool is the core of this project.
That said, the change list for model-explorer-v0.1.33 is not included in the source material, so it cannot be confirmed what differs in this release. There are no benchmarks for performance or accuracy either, and the momentum score is 2.5 out of 10. Its features do fit the job of examining operation graphs of multiple formats as a hierarchy, but more release history needs to accumulate before you can gauge the per-format handling accuracy and the completeness of the feature set.