Model couldn’t finish training because of a configuration or network error in Task Intelligence

[Community Alums]

Hi,

we want to train the model and deploy the model for Case categorization, but we couldn't able to train the model we are getting  Error like " Your model couldn’t finish training because of a configuration or network error. Review the training setup and launch it again, " while train the model. even we have added required roles and enabled properties and added data without empty fields. please help me for this issue.

Thanks!

[DevakarK]

Hi,
Facing issue in Task Intelligence.

[Gsada]

@DevakarK 

Try to check the user "sharedservice.worker" is active, and assign the below roles to train the model

Roles: platform_ml (platform_ml_read, platform_ml_write & platform_ml_create)

if my response helped please mark it correct. Thank you.

Regards,

Dnyandeo

[devakarkesa]

It was already given.

Still Facing the same issue.

[devakarkesa]

Already Given by the system itself while installing plugin.

Still Facing the Same Issue.

[Brian Bakker]

@devakarkesa 

That is not an issue with ML/TI Config as indicated by the previous response, but a data issue, because it connects to the ML Trainer, it was able to download the dataset (this means it is not an ACL/Role issue because it was able to access the data on the instance), create the pipeline for multiple model training (one for each Output field), and then the error occurred during the actual solution training for one of the Output fields. Usually caused by either a data issue with the input fields or the output fields.

Suggestion is to start with one input/output field (i.e. input: short_description, output: assignment_group) and see if that trains successfully. If not, check the data in these two fields on table [incident] based on the condition you provided in the TI categorization solution. If the issue persist, raise a case with ServiceNow Support.