flairNLP
[Question]: Installing `pyab3p`
Question
When trying to run species_linker = EntityMentionLinker.load("species-linker") I am getting 'pyab3p' is not found, switching to a model without abbreviation resolution. This might impact the model performance. To reach full performance, please install pyab3p by running: pip install pyab3p.
When I try to run pip install pyab3p I get ERROR: Failed building wheel for pyab3p due to ab3p_source/Ab3P.cpp:1:10: fatal error: 'Ab3P.h' file not found. Is this a known issue and is there a known fix?
Hello @jessicapetrochuk pyab3p does not work on all operating systems, which is why we included a version that does not require pyab3p.
@alanakbik thanks - how does this impact the performance of the model?
I would say that the impact of abbreviation depends on the entity type, for species specifically it should be negligible, while for instance for disease resolving abbreviations can be critical (think of all the acronyms like "TBC").
If you want to read more about the topic I can recommend these two publications:
- https://aclanthology.org/2023.emnlp-main.893/
- https://academic.oup.com/bioinformatics/article/40/8/btae474/7721929
Hi @jessicapetrochuk
can you please post the full stacktrace/error message/logs? Also can you please specify which python version and OS you are using?
I'm on python 3.11.7 and macOS
Here is the full stack trace:
Collecting pyab3p
Using cached pyab3p-0.1.0.tar.gz (35 kB)
Installing build dependencies ... done
Getting requirements to build wheel ... done
Preparing metadata (pyproject.toml) ... done
Building wheels for collected packages: pyab3p
Building wheel for pyab3p (pyproject.toml) ... error
error: subprocess-exited-with-error
× Building wheel for pyab3p (pyproject.toml) did not run successfully.
│ exit code: 1
╰─> [35 lines of output]
running bdist_wheel
running build
running build_py
creating build
creating build/lib.macosx-13-x86_64-cpython-311
creating build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/__init__.py -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_Lf1chSf.nm -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_Lf1chSf.ha -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/cshset_wrdset3.ct -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_stop.str -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/Lf1chSf -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/Ab3P_prec.dat -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/cshset_wrdset3.str -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/cshset_wrdset3.ad -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_stop.nm -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_stop.ha -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/stop -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_Lf1chSf.ad -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/cshset_wrdset3.ha -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_stop.ad -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/cshset_wrdset3.nm -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/hshset_Lf1chSf.str -> build/lib.macosx-13-x86_64-cpython-311/word_data
copying word_data/SingTermFreq.dat -> build/lib.macosx-13-x86_64-cpython-311/word_data
running build_ext
clang++ -Wsign-compare -Wunreachable-code -fno-common -dynamic -DNDEBUG -g -fwrapv -O3 -Wall -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX13.sdk "-I/Users/jessicapetrochuk/Library/Mobile Documents/com~apple~CloudDocs/Documents/Replifai/replifai/venv/include" -I/usr/local/opt/[email protected]/Frameworks/Python.framework/Versions/3.11/include/python3.11 -c flagcheck.cpp -o flagcheck.o -std=c++17
building 'pyab3p' extension
creating build/temp.macosx-13-x86_64-cpython-311
creating build/temp.macosx-13-x86_64-cpython-311/ab3p_source
clang++ -Wsign-compare -Wunreachable-code -fno-common -dynamic -DNDEBUG -g -fwrapv -O3 -Wall -isysroot /Library/Developer/CommandLineTools/SDKs/MacOSX13.sdk -I/private/var/folders/4j/z95plqpx4m73mjzln4_yjs_00000gn/T/pip-build-env-j5cp525f/overlay/lib/python3.11/site-packages/pybind11/include "-I/Users/jessicapetrochuk/Library/Mobile Documents/com~apple~CloudDocs/Documents/Replifai/replifai/venv/include" -I/usr/local/opt/[email protected]/Frameworks/Python.framework/Versions/3.11/include/python3.11 -c ab3p_source/Ab3P.cpp -o build/temp.macosx-13-x86_64-cpython-311/ab3p_source/Ab3P.o -std=c++17 -mmacosx-version-min=10.14 -fvisibility=hidden -g0 -w
ab3p_source/Ab3P.cpp:1:10: fatal error: 'Ab3P.h' file not found
#include "Ab3P.h"
^~~~~~~~
1 error generated.
error: command '/usr/bin/clang++' failed with exit code 1
[end of output]
note: This error originates from a subprocess, and is likely not a problem with pip.
ERROR: Failed building wheel for pyab3p
Failed to build pyab3p
ERROR: ERROR: Failed to build installable wheels for some pyproject.toml based projects (pyab3p)
Okay, when I built the python-binding, I didn't manage to build for macosx as target, hence it isn't downloading a precompiled version.
The missing "Ab3P.h" file is an error on my side tho. I have just released a new version that contains the missing header-files.
You can try again trying to install pyab3p==0.1.1, however from my experience I would guess that importing pyab3p will cause the python script to crash on macosx.
The pip install and the import both actually work for me after that update - thanks!
Wow that is great, I suppose that means I can close this issue. Feel free to repoen if there is any followup required