Demo - Config validation error
damiansuski opened this issue · comments
I created a new annotation project with cdb an vocabulary based on MedMentions and a small subset of patient notes from pt_notes.csv. When I start a new annotation project a "Config validation error" occurs.
It reoccurs each time I click any document on the list. The exact error messages are
Error: Config validation error ner -> incorrect_spans_key extra fields not permitted {'nlp': <spacy.lang.en.English object at 0x7f2187cd4c10>, 'name': 'ner', 'incorrect_spans_key': None, 'model': {'@architectures': 'spacy.TransitionBasedParser.v2', 'state_type': 'ner', 'extra_state_tokens': False, 'hidden_width': 64, 'maxout_pieces': 2, 'use_upper': True, 'nO': None, 'tok2vec': {'@architectures': 'spacy.Tok2Vec.v2', 'embed': {'@architectures': 'spacy.MultiHashEmbed.v2', 'width': 96, 'attrs': ['NORM', 'PREFIX', 'SUFFIX', 'SHAPE'], 'rows': [5000, 2500, 2500, 2500], 'include_static_vectors': True}, 'encode': {'@architectures': 'spacy.MaxoutWindowEncoder.v2', 'width': 96, 'depth': 4, 'window_size': 1, 'maxout_pieces': 3}}}, 'moves': None, 'update_with_oracle_cut_size': 100, '@factories': 'ner'}
and
Full Error:
Traceback (most recent call last):
File "./api/views.py", line 319, in prepare_documents
CAT_MAP=CAT_MAP, project=project)
File "./api/utils.py", line 301, in get_medcat
cat = CAT(cdb=cdb, config=cdb.config, vocab=vocab)
File "/usr/local/lib/python3.7/site-packages/medcat/cat.py", line 98, in __init__
self._create_pipeline(self.config)
File "/usr/local/lib/python3.7/site-packages/medcat/cat.py", line 105, in _create_pipeline
self.pipe = Pipe(tokenizer=spacy_split_all, config=config)
File "/usr/local/lib/python3.7/site-packages/medcat/pipe.py", line 40, in __init__
self._nlp = spacy.load(config.general['spacy_model'], disable=config.general['spacy_disabled_components'])
File "/usr/local/lib/python3.7/site-packages/spacy/__init__.py", line 51, in load
name, vocab=vocab, disable=disable, exclude=exclude, config=config
File "/usr/local/lib/python3.7/site-packages/spacy/util.py", line 324, in load_model
return load_model_from_package(name, **kwargs)
File "/usr/local/lib/python3.7/site-packages/spacy/util.py", line 357, in load_model_from_package
return cls.load(vocab=vocab, disable=disable, exclude=exclude, config=config)
File "/usr/local/lib/python3.7/site-packages/en_core_web_md/__init__.py", line 10, in load
return load_model_from_init_py(__file__, **overrides)
File "/usr/local/lib/python3.7/site-packages/spacy/util.py", line 523, in load_model_from_init_py
config=config,
File "/usr/local/lib/python3.7/site-packages/spacy/util.py", line 392, in load_model_from_path
nlp = load_model_from_config(config, vocab=vocab, disable=disable, exclude=exclude)
File "/usr/local/lib/python3.7/site-packages/spacy/util.py", line 435, in load_model_from_config
validate=validate,
File "/usr/local/lib/python3.7/site-packages/spacy/language.py", line 1677, in from_config
raw_config=raw_config,
File "/usr/local/lib/python3.7/site-packages/spacy/language.py", line 779, in add_pipe
validate=validate,
File "/usr/local/lib/python3.7/site-packages/spacy/language.py", line 660, in create_pipe
resolved = registry.resolve(cfg, validate=validate)
File "/usr/local/lib/python3.7/site-packages/thinc/config.py", line 730, in resolve
config, schema=schema, overrides=overrides, validate=validate, resolve=True
File "/usr/local/lib/python3.7/site-packages/thinc/config.py", line 779, in _make
config, schema, validate=validate, overrides=overrides, resolve=resolve
File "/usr/local/lib/python3.7/site-packages/thinc/config.py", line 839, in _fill
overrides=overrides,
File "/usr/local/lib/python3.7/site-packages/thinc/config.py", line 901, in _fill
) from None
thinc.config.ConfigValidationError:
Config validation error
There was an issue with the spacy version. In the In Docker container there was a spacy 3.0.8. I forced the installation of the spacy 3.1.3 version with the command (inside container)
pip install spacy==3.1.3
just to see wheather the error disappears. And it disappeared. However, I cannot be sure if it does not mess some dependencies.
hi @damiansuski - thanks for reporting the bug. spacy is getting pulled in via medcat, so you're correct in thinking this might mess with other CDBs.
Did you download the CDB from somewhere or did you create it?
The following order of commands will downgrade spacy to 3.0.8 automatically due to en_core_sci_md not supporting spacy 3.1 but will not downgrade en_core_web_md accordingly.
MedCATtrainer/webapp/Dockerfile
Lines 21 to 23 in 558d58f
That can be confirmed by the docker build logs:
ERROR: medcat 1.2.7 has requirement spacy<3.1.4,>=3.1.0, but you'll have spacy 3.0.8 which is incompatible.
ERROR: en-core-web-md 3.1.0 has requirement spacy<3.2.0,>=3.1.0, but you'll have spacy 3.0.8 which is incompatible.
Installing collected packages: click, typer, spacy, en-core-sci-md
Found existing installation: click 8.0.4
Uninstalling click-8.0.4:
Successfully uninstalled click-8.0.4
Found existing installation: typer 0.4.0
Uninstalling typer-0.4.0:
Successfully uninstalled typer-0.4.0
Found existing installation: spacy 3.1.3
Uninstalling spacy-3.1.3:
Successfully uninstalled spacy-3.1.3
Running setup.py install for en-core-sci-md ... done
Successfully installed click-7.1.2 en-core-sci-md-0.4.0 spacy-3.0.8 typer-0.3.2
Changing the command order and moving line 23 above line 21 should avoid the above downgrading on medcat. Nonetheless, regression tests are needed for finding out if there is any degradation.
Just found en_core_sci_md-0.5.0 requires spacy 3.2. So ultimately, medcat needs to work with spacy 3.2 and I am gonna create a new task on that.
Thanks @baixiac - en_core_sci_md was added in the latest patch build for of our Trust deployments, I wonder if en_core_sci_md-0.5 is backward compatible with 0.4 models...
I am not sure either. Maybe go and ask them to confirm that by creating an issue? https://github.com/allenai/scispacy/issues