Unable to Publish NLU model.

Shikha puri
Tera Contributor
1 ACCEPTED SOLUTION

Hello Shikha,

Sorry for the delayed response, as it has been a busy morning 🙂

I am pleased you managed to resolve the issue using the instructions in the last line of the KB article.

Indeed, you have to repeat the training until it reaches the correct version and after each NLU Model training, delete the [ml_solution] record that it created with the older version number. I will update this KB article, and try to make these steps a little bit clearer to fix the issue.

This situation occurs when you update an existing NLU Model on the target instance from a source instance where the NLU Model has been trained more often, so that the NLU Model versions are out of sync.

I will also provide feedback to the Dev team to see how we can prevent this situation from occurring in the future when transferring NLU Models between instances. We have made further improvements with NLU Model transfers between instances in San Diego, but I have not checked to see if this issue persists with NLU Model versioning.

Best regards,

Brian

View solution in original post

7 REPLIES 7

Hi Brian,

 

It's done and now publish button is visible.

I followed the below line and train it again and again and delete records from ml_solution.list until it comes version 34 then I removed  and the duplicate records with the same Version number (34) and that has active=false OR where the Progress field is empty 

 

Follow that steps until it's come version number 34 ------Repeat the above steps if the Publish button continues to be greyed out, until the Version number for this Solution Name is the highest Version number in this table.

Hello Shikha,

Sorry for the delayed response, as it has been a busy morning 🙂

I am pleased you managed to resolve the issue using the instructions in the last line of the KB article.

Indeed, you have to repeat the training until it reaches the correct version and after each NLU Model training, delete the [ml_solution] record that it created with the older version number. I will update this KB article, and try to make these steps a little bit clearer to fix the issue.

This situation occurs when you update an existing NLU Model on the target instance from a source instance where the NLU Model has been trained more often, so that the NLU Model versions are out of sync.

I will also provide feedback to the Dev team to see how we can prevent this situation from occurring in the future when transferring NLU Models between instances. We have made further improvements with NLU Model transfers between instances in San Diego, but I have not checked to see if this issue persists with NLU Model versioning.

Best regards,

Brian

Ishak
Tera Contributor

Hi @Brian Bakker,

I had a similar case and followed the same steps which enabled me to publish the NLU model. However, another problem occured after publishing the model. I am getting the error if i try to train the model once again :The system encountered an unexpected error and couldn't complete your request. Please contact your system administrator for help or try again". Is this related to the changes I made to solve the previous problem, or does this occur due something else?

 

Best regards,

Ishak