I have got the Intelligent Document Processing(IDP) Application from APPIAN Community and successfully deployed the application by following the instructions.
I have created a new Document Type by following the documentation provided in APPIAN Community.
I have also created a Google Cloud and integrated with IDP Application as per the requirement. But i haven't come across any documentation provided for this on APPIAN Community so i have created as per my knowledge.
Now while training the model with new Document Type by uploading a zip containing 10 sample PDFs and i can see a Process model(DU Train AutoML Natural Language Model) started in the backend and It has created a new DataSet(ID & Name) and begins training the model.
I can see errors have been captured in the process variables(attached screenshot for reference) while the process is still running and finally throws exception after 24 hours.
I feel like i have not created the Google Cloud correctly.
Kindly help me with this.
Discussion posts and replies are publicly visible
Could you run this expression rule in your environment:
and let us know if you get true or false as a result?
I tried executing the mentioned rule and can see TRUE as output.
Can you try uploading the zip again but with at least 20 documents? The docs recommends at least 20:https://docs.appian.com/suite/help/20.1/idp/configuring-idp.html#step-3-upload-example-documents
Please ensure that there aren't duplicate documents amongst the 20, and that none of them are within nested folders.
Even after uploading 20 non duplicate documents, I still see the same error captured while training the model.
Please check your training model details
1) Go to Google Cloud Platform Console->Natural Language->Datasets
2) Check for the status of your dataset
Refer this also
Hi, We are trying to train a new channel and document types and Im getting the exact same geTrainingStatusError, did you ever determine a root cause? Also, via Appian IDP and integration to google do we have direct access to google cloud platform to look at training status directly in google?
For reference I did solve my issue. It turned out the constants for the storage and automl buckets were not set correctly. After being set eveything worked correctly. Figure out the issue by running the DU_getOperationStatusAutoML from designer to see the full response from Google API.
© 2021 Appian. All rights reserved.