webgpu sample using workerd

An example worker using wonnx for image classification on a webgpu-enabled version of workerd.

Upload model to local R2 simulator

This only needs to be done once.

npx wrangler@latest r2 object put model-bucket-dev/opt-squeeze.onnx --local --file models/opt-squeeze.onnx

Launch local development environment

npx wrangler@latest dev

Send request

curl -v -F "file=@images/pelican.jpeg" http://localhost:8787