Remove from __init__

[LuCeHe]

Hi,

I only want to use the dataset, but it crashes since I don't want to install haiku. I have to comment on the clrs.__init__.py the following lines:

# from clrs import models

from clrs._src import algorithms
from clrs._src import clrs_text
# from clrs._src import decoders
# from clrs._src import processors
from clrs._src import specs

from clrs._src.dataset import chunkify
from clrs._src.dataset import CLRSDataset
from clrs._src.dataset import create_chunked_dataset
from clrs._src.dataset import create_dataset
from clrs._src.dataset import get_clrs_folder
from clrs._src.dataset import get_dataset_gcp_url

from clrs._src.evaluation import evaluate
from clrs._src.evaluation import evaluate_hints

from clrs._src.model import Model

from clrs._src.probing import DataPoint
from clrs._src.probing import predecessor_to_cyclic_predecessor_and_first

# from clrs._src.processors import get_processor_factory

and with that change it doesn't complain anymore when I import from clrs._src.clrs_text.huggingface_generators import clrs_generator and produces correctly the dataset. I wonder if it might be desirable for you too to make sure the code doesn't require you to install packages that are irrelevant for the dataset generation.

[PetarV-]

Hi Luca,

Thank you for your kind interest in our dataset!

I'm happy to leave this issue open so others can make advantage of it.
Since the scope of clrs is way more than just its text dataset, we are unlikely to actually make a change like this anytime soon. But it's good to know that there is a workaround that doesn't require the use of Haiku, for those users that do not wish to install it.

Thanks,
Petar