Attention

June 2024 Status Update: Removing DataPipes and DataLoader V2

We are re-focusing the torchdata repo to be an iterative enhancement of torch.utils.data.DataLoader. We do not plan on continuing development or maintaining the [DataPipes] and [DataLoaderV2] solutions, and they will be removed from the torchdata repo. We’ll also be revisiting the DataPipes references in pytorch/pytorch. In release torchdata==0.8.0 (July 2024) they will be marked as deprecated, and in 0.9.0 (Oct 2024) they will be deleted. Existing users are advised to pin to torchdata==0.8.0 or an older version until they are able to migrate away. Subsequent releases will not include DataPipes or DataLoaderV2. Please reach out if you suggestions or comments (please use this issue for feedback)

DataFrameMaker

class torchdata.datapipes.iter.DataFrameMaker(source_dp: IterDataPipe[T_co], dataframe_size: int = 1000, dtype=None, dtype_generator=None, columns: Optional[List[str]] = None, device: str = '')

Takes rows of data, batches a number of them together and creates TorchArrow DataFrames (functional name: dataframe).

Note

There is a trade-off between having a large number of rows within a DataFrame and usage of memory. Please choose a value carefully.

Parameters:

• source_dp – IterDataPipe containing rows of data

• dataframe_size – number of rows of data within each DataFrame, page size can be option

• dtype – specify the TorchArrow dtype for the DataFrame, use torcharrow.dtypes.DType

• dtype_generator – function with no input argument that generates a torcharrow.dtypes.DType, which overrides dtype if both are given. This is useful for when the desired dtype is not serializable.

• columns – List of str that specifies the column names of the DataFrame

• device – specify the device on which the DataFrame will be stored

Example

>>> from torchdata.datapipes.iter import IterableWrapper
>>> import torcharrow.dtypes as dt
>>> source_data = [(i,) for i in range(3)]
>>> source_dp = IterableWrapper(source_data)
>>> DTYPE = dt.Struct([dt.Field("Values", dt.int32)])
>>> df_dp = source_dp.dataframe(dtype=DTYPE)
>>> list(df_dp)[0]
  index    Values
-------  --------
      0         0
      1         1
      2         2
dtype: Struct([Field('Values', int32)]), count: 3, null_count: 0