We don't have a paper to cite yet. I would suggest citing the homepage: https://github.com/onnx/onnx

Can you endorse a common title and author list for the citation? How about this:

ONNX. Open Neural Network Exchange. 2017, https://github.com/onnx/onnx.

How about something like this?

@misc{bai2019,
    author = {Bai, Junjie and Lu, Fang and Zhang, Ke and others},
    title = {ONNX: Open Neural Network Exchange},
    year = {2019},
    publisher = {GitHub},
    journal = {GitHub repository},
    howpublished = {\url{https://github.com/onnx/onnx}},
    commit = {94d238d96e3fb3a7ba34f03c284b9ad3516163be}
}

@bddppq @houseroad @linkerzhang

I'm also interested in how to cite ONNX in a publication. Right now, I'm deferring to @breandan's bibtex above.

@ebarsoum @prasanthpul any thoughts?

@ibrahimhaddad for LF AI recommendation

I think the best proceed ATM would be to "link" this repository to zenodo.org and make it produce citeable references for every release you publish. Then the "ultimate publication" could just reference zenodo reference and improve your h-factor ;)

Some tools which could help along the way for that: https://github.com/all-contributors/ which provides a bot to add contributors from comments in PRs etc.
Also a brand new fresh one (disclaimer: I am related ;)): https://github.com/con/tributors/ which aims to simplify management across all-contributors/.zenodo.json etc.

Since it is Python, I would also advise to look into "integrating" with duecredit: http://duecredit.org/ (again the same disclaimer as to con/tributors) ;)

Has this been resolved? I couldn't find a published paper or preferred citation format. Maybe you could include one at the end of the Readme?

I think that topic should be discussed: Maybe we should use https://citation-file-format.github.io/