Note: This Appian plug-in uses a legacy version of the Google AutoML service, which Google has deprecated. After Google shuts down these services, the Google Cloud AutoML connected system plug-in and the Google AutoML Tables connected system plug-in will no longer function in any version of Appian. Customers who need to continue meeting these use cases will need to contract with Google directly and build an HTTP integration to leverage new services, such as Google Vertex AI.
Overview
Google Cloud AutoML enables building, training, and predicting on custom Image, Text, or Translate models in using Google’s AutoML service. Model details include metadata about the type of model.
Key Features & Functionality
Integrations provided:
Requirements:
I'm using these connected systems with the Appian IDP app. However, when it gets to creating the model (DU_createModelAutoML) it fails with the following error.
Could not perform requested action, verify that the inputs are valid. com.google.api.gax.rpc.FailedPreconditionException: io.grpc.StatusRuntimeException: FAILED_PRECONDITION: To start training the model, it requires each annotation spec has 8 annotations for training, 1 annotations for validation and 1 annotations for testing. Current dataset does not meet the minimum requirement. Please import more data with those annotation specs before training:[2].
I did a quick search and all I found is that the data provided for my training model needs to be annotated but I'm not sure what that means.
The problem was that I was only providing one type of docs which is why it kept complaining because the other type was missing
Hello Jose,
It looks like at least one of the zip files you submitted for training did not contain the minimum of 10 unique sample pdfs. Please reference Step 3 in our configuration documentation - https://docs.appian.com/suite/help/20.2/idp-1/configuring-idp.html.
Thanks