scikit-learn-benchmarks
scikit-learn-benchmarks copied to clipboard
scikit-learn
Repurpose this repo to host asv benchmark results
@scikit-learn/core-devs are you still ok to repurpose this repo (inactive since 2013) to host results from regular runs of the new asv benchmark suite as discussed in a previous core dev meeting ?
If so, what should be done with it's content ? just delete it (it's still in its forks if ever needed) ?
do you mind if I push directly and not use the PR scheme ? it'll make it faster to iterate quickly to set it up.
Also, it will not be my github account who will be pushing the new results here but https://github.com/sklearn-benchmark-bot. Is it ok if I add it as a collaborator (on this repo only) with push permission ?
I'm happy to experiment with the repo directly as you suggest, and I wouldn't worry about the old content since it's gonna be in the git history anyway (I assume you're not intending to delete the history :D )
Thanks for taking care of the benchmarks.
I think it's fine to delete whatever is here and for you to push directly.
Should we rename sciki-learn-speed into scikit-learn-benchmarks? Since we don't benchmark just speed but also memory usage, etc?
Also, it will not be my github account who will be pushing the new results here but @sklearn-benchmark-bot. Is it ok if I add it as a collaborator (on this repo only) with push permission ?
Yes but who's managing @sklearn-benchmark-bot? We should make sure enough core devs have credentials to manage the bot?
Also out of curiosity what's going to happen to https://github.com/scikit-learn-inria-fondation/sklearn-benchmark?
Also out of curiosity what's going to happen to https://github.com/scikit-learn-inria-fondation/sklearn-benchmark?
It will be deleted. It was just to experiment the automation.
Yes but who's managing @sklearn-benchmark-bot? We should make sure enough core devs have credentials to manage the bot?
Good point. Currently only me, but I can give its credentials to you and the other core devs
Should we rename sciki-learn-speed into scikit-learn-benchmarks
I'm ok with that
