This patch release refreshes the package documentation so installation,
model selection, and deployment workflows are easier to discover and follow.
There are no runtime API changes in this release.
Improvements
- Expanded the README to cover the complete workflow, including video,
training, evaluation, ONNX export, pruning, and quantization. - Added a documentation index linking directly to the model, data, training,
evaluation, CLI, and deployment guides. - Split installation instructions into core, ONNX export, and verified
export/quantization paths. - Corrected the ONNX dependency guidance: export requires the
exportextra,
while verification and ONNX quantization also requireonnxruntimeor
onnxruntime-gpu. - Linked the end-to-end optimization example and clarified export compatibility
and reliability guarantees.