Blog post idea: using a separate repo to keep your ML models

[aguschin]

Perhaps this is a point to clarify somewhere. When I think of such tools I personally always imagine to have a separate repo where I store e.g. the mlem models. Why? Mainly because at a certain point I would like to train as part of ci/cd and avoid the creation of new commits to the repo itself as part of the ci/cd. I know it's possible via cml, but I would prefer to avoid the need altogether.

Perhaps it's only me...

Originally posted by @francesco086 in #199 (comment)

🧵 See the thread for more details on this