NLU model stuck in draft. Not changing after publishing.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
10-25-2022 10:21 AM
Hi All,
I added about 10 additional utterances to an NLU intent of an existing Flow and after training and publishing the model without errors it still shows in the 'draft' state.
I tried publishing it again but it still shows 'draft' but active.
Any suggestions will be highly appreciated.
Thanks in advance.

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
11-18-2022 10:33 AM
When this has happened to me in the past, it is because we had a duplicate NLU Model Solution record with the same version number in the ml_solution table; you ran reference KB0952640 if you have access
The fix was to delete the duplicate m_solution (there will be two with the same version number; one will be inactive; delete!)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
12-04-2023 05:26 PM
I'm suuuuper late to the party, but I'm getting the same issue as Lolo but there is no duplicate version in our ml_solution table.

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
12-05-2023 10:10 AM
Does the solution version match the NLU Model published/trained version in table [sys_nlu_model_status]? You can find the solution name in table [sys_nlu_model]. There should also be a [ml_capability_definition_base] record for this Solution Name, and the "Current Version" should also be the same version "9".
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
12-05-2023 11:22 AM
Hi @Brian, I'll have to make a note to check this next time, as we were able to choose another NLU model, save and publish it, then change it back to the intended NLU model and it worked. I think if I were to check your suggestion in our environment now, it would match as we're not seeing the NLU mappings stuck at draft for a particular topic/intent which was changed.