will acuitylite support onnx model + calibration cache format?

VeriSilicon / acuitylite

Acuitylite is an end-to-end neural network deployment tool

https://verisilicon.github.io/acuitylite

MIT License

16 stars 5 forks source link

will acuitylite support onnx model + calibration cache format? #12

Closed zjd1988 closed 6 months ago

lileiigithub commented 1 year ago

Hi, Acuitylite 6.12.0 could support the Onnx model, you can try the newest Acuitylite by the command: pip install acuitylite. And the demo is here https://github.com/VeriSilicon/acuitylite/blob/main/demo/squeezenet_onnx_test.py.

zjd1988 commented 1 year ago

@lileiigithub thanks, actually I want to import quant parameter from QAT

geneyoung commented 6 months ago

AcuityLite can support QAT onnx model with a layer enclosed with Qauntize layer and Dequantize layer，AcuityLite also can limited support some quantze layer such as QLinearConv. Could you please try the newest version and file issue if you meet some problems, it is appreciated if you can attach your part of model so that we can reproduce your problem.