Model Explorer Adapter

The Model Explorer Adapter provides graph extraction and visualization conversion for ML models (TensorFlow SavedModel, GraphDef, FlatBuffer/TFLite, StableHLO/MLIR) to Model Explorer's visualization JSON format.

It is implemented as a stable C-ABI library (libai_edge_model_explorer_adapter.so / .dylib) with a pure Python ctypes FFI wrapper for universal ABI stability across Python versions without pybind11 compile-time matrix bloat.

Install from PyPI

Install ai-edge-model-explorer-adapter via pip from PyPI. For example, in a Python virtual environment:

% python3 -m venv ~/tmp/venv
% source ~/tmp/venv/bin/activate
(venv) $ pip install ai-edge-model-explorer-adapter

Use the Package

After installation, the package can be imported directly:

import ai_edge_model_explorer_adapter as adapter

config = adapter.VisualizeConfig()
json_str = adapter.ConvertFlatbufferDirectlyToJson(config, "model.tflite")
print(json_str)

Build and Install Locally

Declarative Bazel Build

Build the Python wheel directly using Bazel:

bazel build -c opt //python/ai_edge_model_explorer_adapter:wheel

The resulting wheel is located under bazel-bin/: bazel-bin/python/ai_edge_model_explorer_adapter/

Install

Install the resulting wheel via pip:

(venv) $ pip install \
  bazel-bin/python/ai_edge_model_explorer_adapter/*.whl