list' object has no attribute 'to' in trainer function in torch to(self.device)

Question

class Trainer:
    def __init__(self, modelName,model,train_partial_dataset, val_dataset, loss_function, optimizer,path_save_best_model,device):
        
        self.train_dataset = train_dataset 
        self.test_dataset = val_dataset
        self.model = model
        self.modelName=modelName
        self.loss = loss_function 
        self.optimizer = optimizer
        self.device = torch.device( device if torch.cuda.is_available() else 'cpu')
        self.model.to( self.device )
        print('using device: ',self.device)

    def _train(self, train_dataset, num_epochs=1,clip=True ):
  
        train_dataloader = DataLoader( self.train_dataset, batch_size= batch_size, shuffle= False, num_workers= num_dataset_workers)
        val_dataloader = DataLoader( self.test_dataset, batch_size= batch_size, shuffle= False, num_workers= num_dataset_workers)
        #test_dataloader = DataLoader( test_dataset, batch_size= batch_size, shuffle= False, num_workers= num_dataset_workers)
        best_val_f1= -1
        current_best = -1
        best_res = None
        
        torch.cuda.empty_cache()
        start_time = time.time()
        for epoch in range(num_epochs):
          self.model.train()
          train_losses=[]
          print("epoch is ",epoch)
          
          for batch_idx, batch in enumerate(train_dataloader):
            #print(batch)
            
            batch = { k : v.to(self.device) for k,v in batch.items() }    // error statement list v has no function to
            logits,_ = model(batch)
            #resize=batch_dat_function[0].size(0)
            cost = self.loss(logits, batch['targets'].float().to(self.device))
            optimizer.zero_grad()
            cost.backward()
            if clip:
                torch.nn.utils.clip_grad_norm(model.parameters(),0.5)
            optimizer.step()
            train_losses.append(cost.item())
          
          print('train loss is ',np.average(train_losses))
          with torch.set_grad_enabled(False):
            matrix=compute_binary_accuracy(model, val_dataloader, DEVICE)
            curr_val_f1=matrix["f1_score"]
            curr_val_precision=matrix["precision"]
            curr_val_recall=matrix["recall"]
          if curr_val_f1>best_val_f1 :
            best_val_f1=curr_val_f1
            
            print("Saving Best Matrix",matrix)
            counter=0
            torch.save( self.model.state_dict(), path_save_best_model)
            #dill.dump(self.model, open(BestModelDir+ "/"+modelName+"classifier.pkl","wb"))
            SaveBestResult(matrix,ResultfilePath,self.modelName)
          else :
            print("F1-Score has not improved and it is : ",best_val_f1)
          print('Time elapsed: {',(time.time() - start_time)/60 )




if __name__ == "__main__":

    train_dataset = ReviewDataset(Clean_Train_Data, preprocessed= True, vocab= vocab)
    test_dataset = ReviewDataset(Clean_Test_Data, preprocessed= True, vocab= vocab)
    val_dataloader = DataLoader( test_dataset, batch_size= 1, shuffle= False, num_workers= 20)                                                 
    batch_size=128
    num_dataset_workers=20
    
    model = MultiHeadAttBiLSTM(weights, embedding_length, hidden_size, output_size,batch_size)
    path_save_best_model=os.path.join(BestModelDir+ "/"+modelName+".pt")
                                      
    loss_function=F.binary_cross_entropy_with_logits
    optimizer = torch.optim.Adam(model.parameters(), lr=LEARNING_RATE)
    
    
    trainer = Trainer(modelName,model, train_dataset, test_dataset,loss_function, optimizer,path_save_best_model,DEVICE)
    print('dkdbkbdfvbjfdbv')
    trainer._train(train_dataset)

enter image description here

**error statement**
    result_list=result_dataset_generator(val_dataloader)

Inside the trainer class in train function v is considered as list and showing to(self.device) not applicable, what is the mistake in the above code?