Num_workers in DataLoader will increase memory usage?

[ngxbac (Nguyễn Xuân Bắc)]

I have the same problem as him. Here is my topic

Memory fragmentation

Hello, I am dealing with memory fragment when using dataloader. Following is my __getitem__ function: def __getitem__(self, idx): # Get token token = self.tokens[idx] # Get drawing drawing = self.drawings[idx] stroke = stroke_data(drawing).astype(np.float32) # stroke = np.expand_dims(stroke, axis=0) stroke = np.transpose(stroke, (1, 0)).astype(np.float32) drawing = eval(drawing) label = self.labels[idx] # …

Please help us