Hi,
as of today, the clean way with DO in WML with a model in production would be to create a new deployment with the modified model.
However, you can attach in the job you create any file (e.g. by reference from Cloud Object Storage) these files will override (if the case) anything you have initially included in the tarz you deployed. So you could put a new version of a python or OPL model. This is the mechanism that is used to solve LP/MPS problems that are chaging every jobs.
The functionality to cleaning upgrade the model with versioning and lineage is something that is being worked n and should come.
Note that if you are developing/tuning/debugging the model still, not really in production yet, you might look at the DO model builder / experiment in Watson Studio. It allows to easily run different scenarios with different model formulations and/or data and compare. All runs are done on WML underneath in the exact same conditions as you would then in production. So that should be the right way to develop and debug. Then the right final model is deployed to WML.
Hope it helped.
Alain
------------------------------
Alain Chabrier
------------------------------
Original Message:
Sent: Tue June 23, 2020 11:29 AM
From: Jonathan Buttle
Subject: Edit DO model deployed to Watson ML cloud
Hi -
I have a lite account for the Watson machine learning cloud service to run DO models that are too big for DO Community. Is there any way to edit/modify a model that has been deployed or do I have to delete it and start over?
Thanks.
------------------------------
jb
------------------------------
#DecisionOptimization