A microservice used by https://github.com/abragin/book-zipper to match a single chapter.
gunicorn may fail to start because loading transformer model takes too long. To fix it run:
python matching_service/next_match_nn.py
$ gunicorn matching_service.app
...
$ curl -X POST http://localhost:8000/match_chapter -H 'Content-Type: application/json' -d "@sample_request.json"
Cuda is used if available, but can be turned off by setting FORCE_CPU environment variable. If used without cuda huge timeout should be set for gunicorn (e.g. --timeout 240)