ModuleNotFoundError: No module named 'spacy.lemmatizer' on executing fastai's load_learner

[kumuda0]

How to reproduce the behaviour

from fastai.learner import load_learner
classifier = load_learner(os.path.join(path_to_file, "model.pth"))

Error

classifier = load_learner(os.path.join(path, "model.pth"))
File "/usr/local/lib/python3.7/site-packages/fastai/learner.py", line 384, in load_learner
res = torch.load(fname, map_location='cpu' if cpu else None, pickle_module=pickle_module)
File "/usr/local/lib/python3.7/site-packages/torch/serialization.py", line 607, in load
return _load(opened_zipfile, map_location, pickle_module, **pickle_load_args)
File "/usr/local/lib/python3.7/site-packages/torch/serialization.py", line 882, in _load
result = unpickler.load()
File "/usr/local/lib/python3.7/site-packages/torch/serialization.py", line 875, in find_class
return super().find_class(mod_name, name)
ModuleNotFoundError: No module named 'spacy.lemmatizer'

Your Environment

• fastai version: 2.4
• spaCy version: 3.4.0
• Platform: Linux-4.4.0-1128-aws-x86_64-with-debian-9.13
• Python version: 3.7.11
• Pipelines: en_core_web_sm (3.4.0), fr_core_news_sm (3.4.0), es_core_news_sm (3.4.0), de_core_news_sm (3.4.0), it_core_news_sm (3.4.0), en_core_web_trf (3.4.0)

There are suggestions to downgrade the spacy version to fix the issue. But I appreciate a solution with no downgrading of spacy or python.

[polm]

As indicated in this thread, it looks like FastAI is using spaCy v2. Since that's significantly different from spaCy v3, the only alternative to downgrading would be to rewrite the FastAI code to use v3.