This repository has been archived by the owner on Sep 13, 2023. It is now read-only.
0.4.2 🐶
What's new
- Now you can deploy CV models - check out https://iterative.ai/blog/mlem-cv for examples
- Now can deploy models to fly.io with
mlem deploy run flyio
- Added support for sklearn transformers - now MLEM can easily save CountVectorizers, TfidfTransformers, etc
Changes
- Allow to build pip for different python version by @mike0sv in #564
- Return default value for config get if not present by @mike0sv in #563
- Fix empty requirements in notebooks by @mike0sv in #562
- Update GH workflow by @aguschin in #561
- Fix bug with serving models without
model
, with processors only by @aguschin in #569 - Allow rerunning
deploy
without--load
if deployment meta is going to be the same by @aguschin in #570 - Add fly.io deployments by @mike0sv in #511
- Remove empty dirs after artefact cleaning by @mike0sv in #571
- Add pil serializer for images by @mike0sv in #543
- set
--force_overwite True
as default formlem build docker
by @aguschin in #573 - Raise appr error if model option is missing by @mike0sv in #578
- add support for sklearn transformers by @ykasimov in #538
New Contributors
Full Changelog: 0.4.1...0.4.2