Created by: vslyu

Add support for file_list shuffle each epoch This is an example in yaml: phase:

• name: phase1

  model: "{workspace}/model.py" # user-defined model shuffle_filelist: True dataset_name: dataset_train # select dataset by name thread_num: 1

• name: phase2

  model: "{workspace}/model.py" # user-defined model shuffle_filelist: True dataset_name: dataset_infer # select dataset by name thread_num: 1

Fix float learning rate bug learning rate(supported) : float or string write such as 1e-3 ... This is an example in yaml:

• hyper_parameters:

  optimizer: class: SGD learning_rate: 1e-3 / learning_rate: 0.001 ...