Inconsistent filter modules behavior

[zxnie]

Describe the bug

The Score module and Filter module are not behaving the same as ScoreFilter module. They are here: https://github.com/NVIDIA/NeMo-Curator/blob/main/nemo_curator/modules/filter.py

Steps/Code to reproduce bug
from nemo_curator import Score
from nemo_curator.filters import WordCountFilter

"""
Load your dataset as dataset
"""

filter = Score(
WordCountFilter(min_words=80, max_words=200_000),
score_field="word_count",
text_field="text",
score_type=int,
)

filtered_dataset = filter(dataset)

Error message:

Exception: 'TypeError("\'WordCountFilter\' object is not callable")'

Expected behavior

Since all the filters are implemented as a children class of DocumentFilter now (expect for bitext filter), the Score module and Filter module should be consistent with ScoreFilter module and take filter_obj: DocumentFilter as input instead of filter_fn (Callable).

Environment overview (please complete the following information)

• Environment location: local dev env
• Method of NeMo-Curator install: pip install --extra-index-url https://pypi.nvidia.com nemo-curator[cuda12x]
• If method of install is [Docker], provide docker pull & docker run commands used: not applied

Environment details

If NVIDIA docker image is used you don't need to specify these.
Otherwise, please provide:

• OS version: Ubuntu 24.04
• Dask version: 2024.9.0
• Python version: Python 3.10.16 (main, Jan 13 2025, 16:25:23) [GCC 13.3.0] on linux

Additional context

None