mlem
mlem copied to clipboard
iterative
🐶 A tool to package, serve, and deploy any ML model on any platform. Archived to be resurrected one day🤞
We can add caching, pre-collection for known implementations and maybe some other optimizations
Extract `mlem.polydantic` and suggest adding it to `pydantic` itself. It's not clear whether they will accept the change. There is also an issue with openapi schema which probably cannot be...
As discussed with @dmpetrov, as we decided to have a clear separation between functionality of all products, it may be better to have it in DVC only.
Run https://github.com/tonybaloney/perflint
Now it is very simplistic and not informative. What info can we add? Related to #153
Hi 👋 i just discovered MLEM and was intrigued by the deploy options. In the tutorial it is mentioned that one can deploy to Heroku. Since dokku also uses heroku...
Create implementation for Abstract Interface in GTO https://github.com/iterative/gto/issues/46 - Input: artifact path - Output: all details + their text representation There are two use cases for this: 1. API -...
Would be nice to see not only models, but also additional information, like - commit hash which changed the model last time - commit author - commit date - model...
Command should give model description, basically, presenting what was saved in metadata. Something like this: ``` $ mlem describe dog-bark-translator --repo https://github.com/iterative/model-registry --rev main - 📖 Translates dog barks in...
It may work like this (TODO: create more realistic example): ```shell $ mlem versions data/model - main: - 27f5bae aguschin Fixing issue with probabilities - 1de6e4f mike0sv Training on larger...