torchdata.datapipes is not working in Google Colab.
Even after installing the torchdata library, it raises an exception when datapipes function are imported.
I installed the dependencies
!pip install torchdata
or
!pip install --pre torchdata -f https://download.pytorch.org/whl/nightly/cpu/torch_nightly.html
But when I write
import torchdata.datapipes as dp
or
from torchdata.datapipes.iter import StreamReader,FileOpener,FileLister
from torchdata.dataloader2 import DataLoader2, MultiProcessingReadingService
it throws an error :
Exception Traceback (most recent call last)
<ipython-input-15-01a976fbf894> in <cell line: 1>()
----> 1 from torchdata.datapipes.iter import StreamReader,FileOpener,FileLister
2
3 from torchdata.dataloader2 import DataLoader2, MultiProcessingReadingService
6 frames
/usr/local/lib/python3.9/dist-packages/torch/utils/data/datapipes/datapipe.py in register_datapipe_as_function(cls, function_name, cls_to_register, enable_df_api_tracing)
134 def register_datapipe_as_function(cls, function_name, cls_to_register, enable_df_api_tracing=False):
135 if function_name in cls.functions:
--> 136 raise Exception("Unable to add DataPipe function name {} as it is already taken".format(function_name))
137
138 def class_function(cls, enable_df_api_tracing, source_dp, *args, **kwargs):
Exception: Unable to add DataPipe function name sharding_filter as it is already taken
What can be the possible causes and solutions?
The issue is automatically solved now. Library is working fine. Since datapipes are in beta development phase, my guess is that they might have updated some code.