Radim Řehůřek
Radim Řehůřek
Yeah, why not. But either way we should make clear what parameters are mandatory (size_mb) and which are optional, with what defaults.
Not sure what you mean -- can you send a PR?
I vaguely remember discussion around alternative hashing schemes in the beginning. I don't remember the details but my guess is murmur won because of availability, portability & familiarity. Which doesn't...
Billions doesn't strike me as super large. What is the "business cost" of a collisions – how much do collisions / approximate counts matter?
@Daybreak2019 can you open a PR with a fix? Thanks!
Yeah, wheels would be nice. @yaskevich can you help out? CC @mpenkov.
Building wheels is not difficult; the 3rd party tools and services (AppVeyor & co) will be the biggest pain. Hopefully @mpenkov and @menshikh-iv could answer your questions.
There's [gitter](https://gitter.im/RaRe-Technologies/gensim) and there's [twitter](https://twitter.com/gensim_py). But the Gensim [mailing list](https://groups.google.com/forum/#!forum/gensim) and Github here are the liveliest :) What kind of extension are you working on? @mpenkov can you please assist...
Done. Admin now. (You were "maintainer" before – I'd have thought that includes pushing. Weird.)
This is what I see. There's no higher privilege than `admin`: