PyPI - Release kenlm library on PyPI

[patrickvonplaten]

Hello,

Thanks a lot for this amazing library. It seems like more and more speech libraries are moving to Python, e.g.:

• Fairseq: https://github.com/pytorch/fairseq
• Transformers: https://github.com/huggingface/transformers
• Espnet: https://github.com/espnet/espnet
• Speechbrain: https://speechbrain.github.io/
• PyCTCDecode: https://github.com/kensho-technologies/pyctcdecode
• Icefall: https://github.com/k2-fsa/icefall

kenlm is still the standard library for efficient ngram decoding with many of the above libraries have dependencies on it. It's great to have Python bindings, but it would be even nicer if the library could be hosted on https://pypi.org/ to have a clean kenlm dependency on other pypi packages.

@kpu - is there a possibility that this library can be published on pypi?

[kpu]

Yes, I mainly need to learn how to do that.

[patrickvonplaten]

Awesome! Happy to help!

[Narsil]

https://packaging.python.org/en/latest/tutorials/packaging-projects/#generating-distribution-archives

This is my go-to to deploy to pypi !

[sam-writer]

We at Writer have some libraries on PyPi and are happy to help.

@kpu It looks like you own the kenlm name on PyPi and pushed something up in 2017. Do you still have access to that account?

[f4hy]

@kpu Any update on this? The instructions given above should be enough. This would be a great help to be able to use kenlm in standard python pipelines.

[kpu]

https://pypi.org/project/kenlm/0.1/

[patrickvonplaten]

Wuhuuu - this is great! Thanks a lot for the effort @kpu !