Python library for reading ARPA n-gram models.
- Documentation is available.
- Changes between releases are documented.
- Bugs can be reported on the issue tracker.
- Questions can be asked via e-mail.
- Source code is tracked on GitHub.
In order to install the Python 3 version:
$ pip install --user -U arpa
In order to install the Python 2.7 version:
$ pip install --user -U arpa-backport
The package may be imported directly:
import arpa # Python 3.4+
# OR
import arpa_backport as arpa # Python 2.7
models = arpa.loadf("foo.arpa")
lm = models[0] # ARPA files may contain several models.
# probability p(end|in, the)
lm.p("in the end")
lm.log_p("in the end")
# sentence score w/ sentence markers
lm.s("This is the end .")
lm.log_s("This is the end .")
# sentence score w/o sentence markers
lm.s("This is the end .", sos=False, eos=False)
lm.log_s("This is the end .", sos=False, eos=False)
Contributions are welcome!
Write a bug report or send a pull request.
Other contributors have done so before.
Copyright (c) 2015-2018 Stefan Fischer
The source code is available under the MIT License.
See LICENSE for further details.