00000000_dataset.py

from . import BaseDataset, register_dataset

@register_dataset("00000000_dataset")
class MyDataset00000000(BaseDataset):
    """
    TODO:
        create your own dataset here.
        Rename the class name and the file name with your student number
    
    Example:
    - 20218078_dataset.py
        @register_dataset("20218078_dataset")
        class MyDataset20218078(BaseDataset):
            (...)
    """

    def __init__(
        self,
        data_path: str, # data_path should be a path to the processed features
        # ...,
        **kwargs,
    ):
        super().__init__()
        ...
    
    def __getitem__(self, index):
        """
        Note:
            You must return a dictionary here or in collator so that the data loader iterator
            yields samples in the form of python dictionary. For the model inputs, the key should
            match with the argument of the model's forward() method.
            Example:
                class MyDataset(...):
                    ...
                    def __getitem__(self, index):
                        (...)
                        return {"data_key": data, "label": label}
                
                class MyModel(...):
                    ...
                    def forward(self, data_key, **kwargs):
                        (...)
                
        """
        ...
    
    def __len__(self):
        ...

    def collator(self, samples):
        """Merge a list of samples to form a mini-batch.
        
        Args:
            samples (List[dict]): samples to collate
        
        Returns:
            dict: a mini-batch suitable for forwarding with a Model
        
        Note:
            You can use it to make your batch on your own such as outputting padding mask together.
            Otherwise, you don't need to implement this method.
        """
        raise NotImplementedError