Hello,
Can someone please explain me what "Run a Prompt" action expects as a file input?
I've got the prompt tested and working using the template "Reads Receipt". The file is a JPG image of a receipt.
The same image file is used both for testing the prompt and the running the flow.
In the flow, the file is obtained from a MS Forms response.
The file structure is:
- An array variable, like:
{
"name": variables('Anexos')?['Name'],
"contentBytes": variables('Anexos')?['ContentBytes']?['$content']
}
- Onlyt the contenBytes part of the array (base64 encoded or not)
- Or, the path to the OneDrive location of the file using "Get file content using path"
I've searched for a solution here:
But NO GO.
Of the around 600 failed runs I've had so far for this, I get the error:
{"operationStatus":"Error","error":{"type":"Error","code":"InvalidPredictionInput","message":"{\"code\":\"BadRequest\",\"message\":\"Invalid image data.\",\"type\":\"\",\"param\":\"\",\"status\":0,\"innererror\":{\"code\":\"\",\"content_filter_result\":{}}}","properties":{"BackendErrorCode":"InvalidInferenceInput","DependencyHttpStatusCode":"400"},"innerErrors":[{"scope":"Generic","target":null,"code":"InvalidImage","type":"Error","properties":{"MlIssueCode":"InvalidImage"}}]},"predictionId":null}
All I'm trying to do is to is to Run the Prompt in Power Automate. The most basic POC one could think of.
And this is absolutely a waste of time, that could be done so much easier, faster, cheaper with Python, but who knows the insanities upon Power Automate has been built?
Masters, oh Masters, please enlighten me, so I can tell my client Power Automate is so cool and try to enjoy the rest of my life.
THANK YOU!
