Mikhail Sveshnikov

makes sense to do each

1. is done in #340 2. same as before

The second option has an advantage because it should work right now. Though another disadvantage of it is that you will have an extra set of artifacts with basically the...

DVC pipeline will persist the data between stages, so if you have feature engineering and pre-processing as different stages you will have output of them both as files. And for...

Related to #94

This is about optimizing requirements collection from live objects: caching, pre-collection and stuff

For now we don't have performance issues with this

Duplicate #34 ?

this issue is about collecting exact versions of requirements - sometimes there are conflicts (eg with other package dependencies). This is rare, so low priority

Probable not public release?