* fix model name

* fix: bug when distillation

* modify some default hyperparams to adapt to fine-tune downstream tasks

1. unset EMA because of the relatively small size of most downstream dataset;
2. use mean and std of IMN.

* update new_user/ShuffleNetV2_x0_25.yaml

* add support for infer

* add export model support

* fix yaml

* fix post process name

* fix topk name

* add support for train and eval

* rm unsed code

* add support for metric save and load ckp

* polish trainer