Is the MLFlow platform the only platform for reusable ML? Is it necessarily the best?

Asked Jul 01 '23 at 06:17

Active Jul 01 '23 at 06:17

Viewed 12 times

Is the MLFlow platform the only platform for reusable ML?

Is it necessarily the best (yes it can be subjective and case-dependent)?

I found at least one paper that discusses this platform:

Accelerating the Machine Learning Lifecycle with MLflow Matei Zaharia, Andrew Chen, Aaron Davidson, Ali Ghodsi, Sue Ann Hong, Andy Konwinski, Siddharth Murching, Tomas Nykodym, Paul Ogilvie, Mani Parkhe, Fen Xie, Corey Zumar Databricks Inc.

https://www-cs.stanford.edu/people/matei/papers/2018/ieee_mlflow.pdf

But I think the platform sounds like it promises a lot. Reusability is a desired property for research, but it can also be a goal of corporate marketing endeavors. The site says it's open source, but there are also open source things that are not goodwilling.

asked Jul 01 '23 at 06:17

mavavilj

I'm at Databricks, so do recommend MLflow, however it's not the only tool for this. Weights and Biases is also commonly used. It pretty well does what it says, it's open source, that much is real. – Sean Owen Jul 01 '23 at 15:18
@SeanOwen Some source claimed Apache Airflow is an alternative. – mavavilj Jul 01 '23 at 17:27
Airflow? no, really not an ML tool at all – Sean Owen Jul 06 '23 at 19:32

Is the MLFlow platform the only platform for reusable ML? Is it necessarily the best?

0 Answers0