Jules S. Damji

Yes, the user guide, as a reference implementation of a data source, will provide immense value for third-party engineering teams who wish to integrate with Ray Data for data ingestion...

@woop Yes, it's a matter of striking a balance and compromise: support a data source out of the box that's has wider adoption, reliable vendor support, and large user or...

@amesar Thanks for filing this. Yeah, this is an interesting enhancement as it poses a question about how long should we retain experiment runs and associated artifacts for auditing purposes....

@Dhaval08 thanks for filing the feature enhancement. What would you prefer: live tail of the logs in the UI or after the run to inspect the logs?

@Dhaval08 I wanted to circle back and clarify some functionality delineation that might be misconstrued. Or we elected the wrong phrases in the documentation that might have easily misinterpreted. MLflow...

@hughperkins Thanks for filling this. Were you thinking along the lines of supporting an OR clause in the `filter_string = "tags.s.release ILIKE '%rc%' or metric.m > 2.5"` in the [mlflow.search_runs()](https://mlflow.org/docs/latest/python_api/mlflow.html#mlflow.search_runs)...

@abhiksark Note that PyTorch autolog only works for [PyTorch Lightning models](https://mlflow.org/docs/latest/python_api/mlflow.pytorch.html#mlflow.pytorch.autolog), not for all PyTorch models. So no surprise that it does not work for general PyTorch models. So this...

cc: @harupy Do you know if this is on the roadmap?