This code will be updated after Oct.

The quantization part is given here

-------------------------------- auto quan----------------------------

rcq.quan.utils

find_modules_to_quantize() : Conv2d -> QuanConv2d , Linear QuanLinear

QuanConv2d & QuanLinear rewrite in quan.func

replace model

args_quan = get_config(default_file='/code/mmclassification/rcq/config.yaml')

modules_to_replace = utils.find_modules_to_quantize(model, args_quan.quan,i_)

model = utils.replace_module_by_names(model,modules_to_replace)