How to use MLFlow models with managed endpoints and private pip

Question

question

ThomasH-1455 asked Apr 21, '22 ThomasH-1455 edited May 24, '22

How do you create a batch endpoint using an MLFlow with a private package installed via a private pip?

I have tried multiple ways (see this github issue: https://github.com/Azure/azure-sdk-for-python/issues/22441#issuecomment-1104082706) but I'm not really getting anywhere.

azure-machine-learning azure-machine-learning-inference

· 4

5 |1600 characters needed characters left characters exceeded

Attachments: Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

YutongTie-MSFT · Apr 22 at 03:57 AM

Thanks for reaching out to us, we are investigating internally and will let you know soon.

Regards,
Yutong

0 ·

ThomasH-1455 YutongTie-MSFT · May 10 at 02:38 PM

Any progress?

Regards.
Thomas

0 ·

ThomasH-1455 ThomasH-1455 · May 16 at 09:06 AM

Ping @ramr-msft :-)

0 ·

ThomasH-1455 YutongTie-MSFT · May 24 at 11:37 AM

0 ·

Answer 1 · 2022-05-02T16:04:56.39Z

Jaya-0415 answered May 2, '22

Is this doable? I explored several times but all failed.

Can Microsoft add documentation about that?

5 |1600 characters needed characters left characters exceeded

Attachments: Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.