search

castorini / pygaggle

a gaggle of deep neural architectures for text ranking and question answering, designed for Pyserini
http://pygaggle.ai/
Apache License 2.0
338 stars 98 forks source link

File "pipes.pyx", line 1465, in spacy.pipeline.pipes.Sentencizer.from_nlp TypeError: type() takes 1 or 3 arguments #246

Open hadifar opened 2 years ago

hadifar commented 2 years ago

Running the re-ranking example throws the following exception:

Traceback (most recent call last):
  File "/project_scratch/t5_ranking.py", line 2, in <module>
    from pygaggle.rerank.transformer import MonoBERT
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/rerank/transformer.py", line 13, in <module>
    from .similarity import SimilarityMatrixProvider
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/rerank/similarity.py", line 5, in <module>
    from pygaggle.model.encode import SingleEncoderOutput
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/model/__init__.py", line 6, in <module>
    from .encode import *
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/model/encode.py", line 8, in <module>
    from .tokenize import BatchTokenizer
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/model/tokenize.py", line 172, in <module>
    class SpacySenticizer:
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/pygaggle/model/tokenize.py", line 174, in SpacySenticizer
    nlp.add_pipe(nlp.create_pipe('sentencizer'))
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/spacy/language.py", line 302, in create_pipe
    return factory(self, **config)
  File "/project_scratch/users/amir/projects/RAG/dialdoc-task/env/lib/python3.9/site-packages/spacy/language.py", line 1045, in factory
    return obj.from_nlp(nlp, **cfg)
  File "pipes.pyx", line 1465, in spacy.pipeline.pipes.Sentencizer.from_nlp
TypeError: type() takes 1 or 3 arguments

Do you know what's the problem and how can I fix it?

spacy version: 2.2.4 tokenizers version: 0.10.3 transformers version: 4.15.0

Python 3.9.7

Thanks

ronakice commented 2 years ago

Install from source? The spacy version is not 2.2.4 in the requirements file

hadifar commented 2 years ago

Thanks for your quick reply.

I installed the library by running pip install -r requirements. The issue was due to my python version, the problem fixed when I switched to python 3.7 :)

lintool commented 2 years ago

Hi @hadifar - Can you try using Python 3.8 and see what happens? cf. https://github.com/castorini/pyserini/issues/916