Method load

tensorpack/dataflow/serialize.py:108–115

Note: If you found deserialization being the bottleneck, you can use :class:`LMDBData` as the reader and run deserialization as a mapper in parallel.

(path, shuffle=True)

    @staticmethod
    def load(path, shuffle=True):
        """
        Note:
            If you found deserialization being the bottleneck, you can use :class:`LMDBData` as the reader
            and run deserialization as a mapper in parallel.
        """
        df = LMDBData(path, shuffle=shuffle)
        return MapData(df, LMDBSerializer._deserialize_lmdb)

    @staticmethod
    def _deserialize_lmdb(dp):