Closed zhixiongzh closed 2 years ago
prepare_by_platform
for now.
prepare_by_platform
for now.
Thank you for the quick response. In this case, PTQ and QAT use the same function to prepare the model, then how does the function tell it should be a PTQ or QAT? Only by training or not in the next step?
If possible, can I get a complete example to see how QAT works? I am very interested in the tool, but it seems most of the configuration parameters are not provided/demonstrated.
For example, in prepare_custom_config_dict
, I can only find out how to set extra_qconfig_dict
, but no idea and no document to demonstrate what the parameters in extra_quantizer_dict
, concrete_args
, and extra_fuse_dict
look like.
You can find few answers here https://mqbench.readthedocs.io/en/latest/. We notice the lack of details , those can be found in source code easily actually.
Hi,
I can't find the definition of prepare_qat_fx_by_platform, which should be in the library mqbench.prepare_by_platform.
Do you just forget to upload the function?