ray.data.datasource.BlockBasedFileDatasink.__init__#

BlockBasedFileDatasink.__init__(path: str, *, min_rows_per_file: int | None = None, **file_datasink_kwargs)[source]#

Initialize this block-based file datasink.

Parameters:
  • path – The folder to write files to.

  • min_rows_per_file – The target minimum number of rows per file. When None, rows are not buffered before being written.

  • **file_datasink_kwargs – Additional keyword arguments forwarded to _FileDatasink.