NeutralizingBias

PID

training process

1. please check these codes are available:

   1. app.py 76 line: trainer.train()

      # app.py  76 line
      
          trainer.train()

   2. model.py 189 line: self.items = ......

      # model.py  189 line
      
              self.items = torch.cat([self.items_id, self.item_pop_true.cuda()], dim=1).to(torch.device('cuda:0'))

   3. utils.py 52 line: date_time = 'the unique name for the model'

      # utils.py  52 line
      
              date_time = '2022' 

2. please check these codes are not available:

   1. model.py 184 line - 186 line

      # model.py   184 line - 186 line
      
              # item_pop = np.zeros((self.num_items, 1))
              # self.item_pop = Parameter(torch.from_numpy(item_pop).float()).requires_grad_(requires_grad=False)
              # self.items = torch.cat([self.items_id, self.item_pop.cuda()], dim=1).to(torch.device('cuda:0'))

   2. trainer.py 71 line: self.max_epoch ...

      # trainer.py  71 line
      
              # self.max_epoch = 48

3. run

   python -u app.py

testing process

1. please check these codes are available:

   1. model.py 184 line - 186 line

      # model.py   184 line - 186 line
      
              item_pop = np.zeros((self.num_items, 1))
              self.item_pop = Parameter(torch.from_numpy(item_pop).float()).requires_grad_(requires_grad=False)
              self.items = torch.cat([self.items_id, self.item_pop.cuda()], dim=1).to(torch.device('cuda:0'))

2. please check these codes are not available:

   1. app.py 76 line: trainer.train()

      # app.py  76 line
      
          # trainer.train()

   2. model.py 189 line: self.items = ......

      # model.py  189 line
      
              # self.items = torch.cat([self.items_id, self.item_pop_true.cuda()], dim=1).to(torch.device('cuda:0'))

3. please note that you should :

   1. utils.py 52 line: date_time = 'the unique name for the model'

      # utils.py  52 line
      
              date_time = '2022' 

   2. trainer.py 71 line: self.max_epoch = the best epoch

      # trainer.py  71 line
      
              self.max_epoch = 48

4. run

   python -u app.py

IID

training & testing process

python -u app.py