News for hackers. Stuff that matters.
Re:Differences between the word2vec paper and its implementa... by liuliu on Monday June 03, 2019 @06:55PM (#20089825)
Maintaining a pipeline that works takes effort that can be spent somewhere else. As long as it can be reproduced with a little effort from me, I really don't mind. I would rather they spend time to try a lot of different things, build intuitions, and search better backbone structures than maintaining a CI system so that every time their code updates can train imageNet to < 20% top-1 error rate.

That should be left to TensorFlow model zoo or GluonCV model zoo. I simply look at these research "open-source" as reproducible research.

Re:Differences between the word2vec paper and its implementa... by billconan on Monday June 03, 2019 @07:04PM (#20089894)
I'm not asking for much, but the code should be runnable at least.

This is why I think jupyter notebook and things alike are so important as the future publishing media. Reproducibility is very important.

Re:Differences between the word2vec paper and its implementa... by taeric on Tuesday June 04, 2019 @09:50AM (#20094978)
Jupyter notebooks are laughably not reproducible right now. They may get there, but they encourage bad development habits. Actively.
Re:Differences between the word2vec paper and its implementa... by PaulHoule on Monday June 03, 2019 @06:56PM (#20089828)
Yeah but I wonder what fraction of CS papers test the algorithm that they thought they were testing...
Re:Differences between the word2vec paper and its implementa... by seanmcdirmid on Monday June 03, 2019 @09:33PM (#20090904)
A lot of CS papers aren’t testing algorithms at all.
Re:Differences between the word2vec paper and its implementa... by opportune on Tuesday June 04, 2019 @12:59AM (#20092045)
It is very common in the last few pages of a CS paper to benchmark your results against other algorithms or models
Re:Differences between the word2vec paper and its implementa... by seanmcdirmid on Tuesday June 04, 2019 @08:49AM (#20094428)
And it’s also very common not to. CS papers are pretty diverse in what they are talking about, many aren’t algorithm papers.