ml flow ui can't be accessed from google colab

[divyag11]

i can't access mlflow ui from google colab,however through jupytre notebook,it's accessible.

[pazlvbanke]

@divyag11 I can

[tienthegainz]

@divyag11 I can

How do you do that?
mlflow ui expose in 127.0.0.1 means that it can't expose to the outsider, change host to 0.0.0.0 may work IDK.

[pazlvbanke]

@tienthegainz run ‘’’mlflow ui -h 0.0.0.0’’’ on you local machine, access by localhost:5000

[tienthegainz]

@tienthegainz run ‘’’mlflow ui -h 0.0.0.0’’’ on you local machine, access by localhost:5000

The question is asking how to read it from Google Colab not your local machine.

[dmatrix]

I would not consider this as a bug.

[dmatrix]

Here is an example

I would not consider this a bug. Can we close this issue?

[sp7412]

So there's no way to make this work in the cloud?

[dmatrix]

So there's no way to make this work in the cloud?

No, at this moment MLflow UI works only in the Databricks and Azure cloud.

@tienthegainz run ‘’’mlflow ui -h 0.0.0.0’’’ on you local machine, access by localhost:5000

The question is asking how to read it from Google Colab not your local machine.

Currently, as far as I know, you can't. But if anyone in the community can make it work, we would love to hear and make a contribution.

[dmatrix]

I have updated the notebook to show how to use MLflow in Google Colab. You can use pyngrok to tunnel. Realy simple!
https://github.com/dmatrix/google-colab/blob/master/mlflow_issue_2350.ipynb

[desertnaut]

@dmatrix many thanks for this!

Since this issue still appears open and your post is rather "buried" in here, I took the liberty to post your code as an answer to the Stack Overflow question Setting-up MLflow on Google Colab, with full attribution of course (including links to your notebook and here).

In case you object to that reposting, please take a minute to inform me here, and I will remove the post immediately.

Thanks again!

[dmatrix]

Am closing this since we have a solution now.