ONNX version of the model
#8
by asofter - opened
Converting the model to ONNX for faster inference with optimum from Huggingface
Thanks for this, I've confirmed that it works for our purposes using candle_onnx (which supports all needed ops right now).