Hoping someone can explain or breakdown an error I am receiving.
I am working off an AI Model to run through Automate to extract information from a document and if confidence score of that document is lower than 70% to kick it to the AI Builder Feedback Loop.
Which according to the action description kicks it back to AI Builder so that I can review and re-train the AI model.
Under the Feedback loop section I have the following condition:
Feedback Loop Variable is activated & Collection Confidence Score is less than 70%
I tested the flow with a totally unrelated document thinking it would fail the AI Model, and kick to the Feedback review.
Except I keep getting an error on the Feedback Loop Section related to that Condition Sample.
Types within separated conditions are independent, meaning that the variable is tested against the Yes string and the collection confidence score is tested against 0.7.
You can try by removing the condition on the feedback loop variable and just keep the one on the confidence score.
I tried this condition on my side with the Collection confidence score and it works:
I've tried .7 previously and had no luck.
It seems that an earlier step in the process creates a variable "Feedback Loop Activated" that is later activated in the condition sample. And that is expecting a string "yes" which it gets in the later condition sample but also a 'float' 70% which gives off the error. I can change the activation type to float or other types but confused what to change it to, to process as intended which is look at confidence score and send for review if not over the threshold.
Hi @raekwon231 ,
It seems that your condition fails because types that don't match in the comparison (between float and string). Try replacing "70%" by "0.7", that may solve the issue.