ort-toolkit

Download a model

on Linux
set https_proxy to proxy-us
pip3 install onnx onnxruntime
git clone https://github.com/xenova/transformers.js/
download a model via for example "python -m scripts.convert --model_id apple/mobilevit-small"

Comment out "SHELL:-s MAXIMUM_MEMORY=4294967296" in onnxruntime/cmake/onnxruntime_webassembly.cmake
Run browser with "--enable-experimental-web-platform-features --enable-blink-test-features --enable-features=WebnnMojoContext --enable-machine-learning-neural-network-api"
Gaps. Browser implementation: computeSync and buildSync, more ops

Name		Name	Last commit message	Last commit date
Latest commit History 26 Commits
.gitignore		.gitignore
README.md		README.md
index.html		index.html
models.js		models.js
sortable.min.css		sortable.min.css
sortable.min.js		sortable.min.js
util.js		util.js