A PyTorch implementation of various hidden Markov model inference and learning algorithms. The library's primary purpose is to provide a means of fitting hidden Markov models using GPU hardware. The library provides a modular interface for fitting HMM's with custom emission models; it has built-in emission models for discrete and Gaussian outputs. The library also supports regularization, the ability to fit the model with multiple random restarts, and supports PyTorch's packed list input represention (so sequences can having varying length).
API documentation can be found at: https://chris.maclellan.hq-git.soartech.com/TorCHmM If this link is unavailable, then the documentation can be built locally. To build a local copy of the documentation, go to the docs folder and run the commands pip install -r doc-requirements.txt and make html to build an HTML copy of the docs. These docs can then be accessed locadlly at _build/html/.
The package can be installed via pip directly from git. To do this run the following command:
pip install -U git+https://<GIT URL>@master
substitute the appropriate git url in the command above.
Once the package has been installed, examples of how to use it can be found in the benchmarks folder.
DISTRIBUTION STATEMENT A. Approved for public release: distribution unlimited.