
Please consider the following steps to get the results of the tables.

1. install the python environment by 
  ``$ pip3 install -r requirements.txt ''

2. download the data sets from the links that we support in the paper.

3. process the data sets to the pickle format. (We give an example in ``./datasets/pickle-format/eur-lex.pkl'')

4. execute train.sh 

5. execute predict.sh (We support the predict logs of four data sets in ``./finished-logs'')

6. use the command
    ``$ tail -n3 ./finished-logs/* > all_log''
  to process the predict log. 

7. execute process_log.py to output mean and std

