iterative
Support "unknown" model type
For unsupported model types we can add a "mock" model_type.
This way we'll enable users to use different mlem commands that dont need to load actual model like clone.
RN, the process is python model obj -> getting model type with analyzer (methods, requirements, io)-> producing artifacts from ModelIO -> saving metadata
For unknown models, users can do analyzers work manually like this
python model object -> user saves model manually -> generates empty metadata from files like ModelMeta.from_files(...) -> optionally fills metadata manually (adding requirements and stuff) -> saves metadata
@mike0sv, is this still an issue? This related to that discord question we had recently.
It's more of a feature proposal than an issue. Don't think it's relevant to that question either, it's basically about an API to manually create MlemModel objects more easily. Rn you can also do this of course, but in very unfriendly way