Answered

Is it a AI Builder Limitation Or Flow Limitation

(1) Share

Report

Posted on by Suga

Hi Folks, I have built a model (494 fields) and trained with 5 sample documents. when I was trying to use this with Power automate, I'm getting the below error message

Request to XRM API failed with server error: 'Message: Flow server error returned with status code "InternalServerError" and details "{"error":{"code":"DynamicSchemaResponseTooLarge","message":"The dynamic schema response from API 'commondataserviceforapps' operation 'GetPredictionSchema' is too large, only schemas with at most '1024' properties are supported."}}". Code: 0x80097376 InnerError: '. The tracking Id is '3e97a6ce-c295-448a-9984-db6ff0ae2e4c'.

How to workaround this limitation Issue or solve it ?

Categories:

AI Builder

I have the same question (0)

All responses (10)

Answers (1)

Sort by

Mira Ghaly 11,415 Moderator on at

Like (0)

Report
Copy link

Link copied!

Hello @JoeF-MSFT would you able to help here or shed some light? Or a possible solution or workaround?

Was this reply helpful? Yes No
Verified answer

JoeF-MSFT Microsoft Employee on at

Like (1)

Report
Copy link

Link copied!

Thanks for tagging me @Mira_Ghaly. Hi @Suga - currently, there is a limit of 300 fields for document processing models to be able to use it in a flow, and while we do have plans to increase this limit, we do not have a specific date for when this will happen.

I can understand your frustration, especially after tagging almost 500 fields. However, I have a workaround that you can use to implement your model in a flow. You can refer to the following guide: Document processing models with more than 300 fields workaround.pdf (windows.net) it requires extra work but can unblock you until we increase the size limit for the standard AI Builder action.

Was this reply helpful? Yes No
Mira Ghaly 11,415 Moderator on at

Like (0)

Report
Copy link

Link copied!

@JoeF-MSFT

Appreciate your response is this limit mentioned anywhere in the Microsoft Docs?

Thanks

Was this reply helpful? Yes No
Mira Ghaly 11,415 Moderator on at

Like (1)

Report
Copy link

Link copied!

@JoeF-MSFT Thanks this workaround worked for us, Very much appreciated!

Was this reply helpful? Yes No
Suga 6 on at

Like (0)

Report
Copy link

Link copied!

Thanks @JoeF-MSFT and @Mira_Ghaly

Was this reply helpful? Yes No
JoeHaslam 4 on at

Like (0)

Report
Copy link

Link copied!

@JoeF-MSFT Do you have anything documented on this limitation? I am getting the same error on a Flow i'm creating which is using my Ai Model. The model has 188 text fields, 1 table and 59 checkboxes. This should be under the 300 items limit as far as I can tell, so I'm not sure why I am getting the error.

I can't use the workaround in my organisation.

Many thanks

Joe

Was this reply helpful? Yes No
plarrue Microsoft Employee on at

Like (0)

Report
Copy link

Link copied!

Hi @JoeHaslam ,

You can find the limitation in this doc Requirements and limitations for document processing models - AI Builder | Microsoft Learn

Thanks

Was this reply helpful? Yes No
JoeHaslam 4 on at

Like (0)

Report
Copy link

Link copied!

Thanks for getting back to me @plarrue ,

My model doesn't have 300 items but I did some testing this morning and removed the table which is 12 rows and 4 colums and the flow saved successfully. Does this mean that every field or block in a table is counted rather than the table as a whole? I'm still not sure why it's failing as my counting gets me to 291 with the table

188 text field items
59 Check box items
1 Table - 44 blocks not including the 4 column headers - and i've set it to ignore the first row which in our form is an example of what inforatmion they're looking for.

so this comes to 291 or 295 including the 4 headers or 299 if it includes the ignored first row.

Any help to explain this would be appreciated.

Many thanks

Joe

Was this reply helpful? Yes No
JeffKing 7 on at

Like (0)

Report
Copy link

Link copied!

I have 77 fields, 16 tables and 17 checkboxes and am getting this same error when I try to use my model in a flow. Ran across this solution and tried to implement it, but I'm getting an error message that it's a bad request. Any idea what the trouble may be?

Was this reply helpful? Yes No
JeffKing 7 on at

Like (0)

Report
Copy link

Link copied!

There are issues with the suggested workaround. In your pdf you have the "Perform an unbound action" highlighted, but state that we should select the "Perform a bound action" instead. Also, my options different: The action in your example shows a purple database icon for the action, but mine remains the green dataverse icon. The parameter "request" in your example is shown as "item/request" in mine. Also, when I paste the prediction response from the generated json, into the parse json action, it doesn't provide me with the button "Generate from sample" - mine shows "Use sample payload to generate sample". Is this no longer a viable solution?

Was this reply helpful? Yes No