You’re offline. This is a read only version of the page.
22
My flow is set up to extract the data from a form and input it into a new row in excel. This is triggered by a specific email. When testing the flow I have found that the flow works as intended however I get an error message stating "Action 'Extract_information_from_documents' failed" and an extra row is added to the table with no data.
The error details box states:
"{"operationStatus":"Error","error":{"type":"Error","code":"InvalidPredictionInput","message":"Encountered an issue when reaching Form Recognizer: InvalidRequest","properties":{"BackendErrorCode":"InvalidInferenceInput","DependencyHttpStatusCode":"400"},"innerErrors":[{"scope":"Record","target":null,"code":"RecordDimensionsOutOfRange","type":"Error","properties":{"ErrorCode":"InvalidRequest","MlIssueCode":"RecordDimensionsOutOfRange"}}]},"predictionId":null}"
I'm not sure what is causing this error as my excel file populates as intended and by all measures it appears the flow is in fact working.
It shows there being 3 apply to each processes, process 1/3 and process 2/3 work correctly. However at process 3/3 the error message is generated. I don't want process 2/3 to add a line in the excel sheet either although currently it is.
1/3, it works as intended and I thought this would be the only process running.
2/3 works but I don't want it to add a row in the excel sheet and don't see what this process is actually doing or needed for.
The error message comes from 3/3 as seen below
I'm assuming the error has to do with my apply to each function, perhaps I could use conditionals to avoid the issue? If anyone has any info that could help me out it would be much appreciated!
I really appreciate the help! The AI Model has 99% accuracy and there is never going to be a time where any of the fields are left blank. When I was running the test all fields in the PDF attached to the email were filled out entirely.
All the correct information goes into my excel table as desired, the problem is that another row is added with no info (only one row needs to be added and it contains the information I need for excel) and then after this the flow "Fails" if I could simply make the flow stop after the first row of information is added to excel it would work exactly how I want.
I don't see the operational status field within the AI, I think my problem should be an easy fix within the flow just not sure what exactly I need to change.
Thanks for posting. Understood. OK, I can't fix the empty value because it is inside your AI model and can't be addressed with a conditional statement (now that I see the flow). My recommendation on the error is to revisit the AI model and verify the operational status field has a value (the error says it is blank). If it is there, then retrain your model to accurately capture this. If you have a few documents where this is going to be blank, include those in your training model so your AI extraction step understands the blanks.
On the 2 of 3 where you don't want to write that to Excel, find it in your Excel step and simply remove the value so it doesn't write it. Here's an example of that: if you didn't want to write the City to your Excel table, find City in the Excel step and remove the dynamic extracted value and leave it blank.
Please let me know if that was helpful! Always glad to help! Tom
Thanks so much for the help! I have attached screenshots of the flow with the steps expanded, if there is a better way for you to see the flow let me know and I can try and do it a different way. Thanks again!
Hi @kellen ,
Sure can. Please add your flow in this thread with the steps expanded so I can see what is being selected. From there, I'm going to add a condition that says if the value you are extracting is not blank, do this, otherwise do nothing - in the apply to each. Always glad to help! Tom
Hi @kellen ,
I would recommend you put your extract info action in a conditional step and check if empty() is equal to true(). Please mark as a solution if that works for you! Always glad to help! Tom
#1
WarrenBelz
146,601
Most Valuable Professional
#2
RandyHayes
76,287
Super User 2024 Season 1
#3
Pstork1
65,942
Most Valuable Professional
Share
Report
All responses (
Answers (
