[GitHub] [beam] damccorm commented on issue #22572: [Feature Request]: Allow specification of a custom model inference method for a RunInference ModelHandler

[GitBox <gi...@apache.org>]

damccorm commented on issue #22572:
URL: https://github.com/apache/beam/issues/22572#issuecomment-1235374438

   Generally, my take here is that we should do option 3 here and allow users to pass in a custom function. Basically:
   
   1) This is a reasonably common pattern
   2) Adding support shouldn't be too hard
   3) Asking users to create their own handler (or model wrapper) any time they want to use a different method doesn't scale well. Some might contribute it back to the community, most won't, and even with those who do we're incurring an extra review/maintenance burden. It also significantly raises the bar for first time users who now would need to understand the handler internals.
   4) Supporting this doesn't meaningfully make it harder for users in the simple use case (omitting this param should do nothing).
   
   @jrmccluskey could you pick this one up when you have the bandwidth.


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: github-unsubscribe@beam.apache.org

For queries about this service, please contact Infrastructure at:
users@infra.apache.org