This browser is no longer supported.
Upgrade to Microsoft Edge to take advantage of the latest features, security updates, and technical support.
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(172.8, 78%, 26%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EM%3C/text%3E%3C/svg%3E)
As titled. I have been training a custom model using document intelligence, and recently looking into composing multiple models. However I am unable to figure out a way to select which labeled data I want to use to train my models (I'm also unsure if there's a way to do so). If there isn't such a way to do so, then how should I train a compose model and separate the data out?
Any sort of help/response would be greatly appreciate. Thanks in advance!
![](data:image/svg+xml, %3Csvg xmlns='http://www.w3.org/2000/svg' height='64' class='font-weight-bold' style='font: 600 30.11764705882353px "SegoeUI", Arial' width='64'%3E%3Ccircle fill='hsl(195.2, 4%, 28%)' cx='32' cy='32' r='32' /%3E%3Ctext x='50%25' y='55%25' dominant-baseline='middle' text-anchor='middle' fill='%23FFF' %3EVM%3C/text%3E%3C/svg%3E)
Hello @Mason , Thanks for using Microsoft Q&A Platform.
Generally, when you train a custom model, you can label the data to help the model learn to recognize the fields you are interested in.
While creating a composed model, you can assign custom models that you've already created to a single composed model ID. To analyze a document with a composed model, Document intelligence first classifies the submitted form, chooses the best-matching assigned model, and returns results.
As mentioned, you need to provide the trained custom model ID into the composed model, no additional labeled data is required.
To Compose document model, here is the REST API sample request:
POST https://myendpoint.cognitiveservices.azure.com/documentintelligence/documentModels:compose?api-version=2024-02-29-preview
{
"modelId": "composedModel",
"description": "Composed model description",
"componentModels": [
{
"modelId": "model1"
},
{
"modelId": "model2"
}
]
}
Is this something you are looking for?
Hey @VasaviLankipalle-MSFT , sorry but I think I didn't state my question clear enough, so let me try explaining with an example. If I have a set of data A and a set of data B that I want to use to train two different models and put them into a composed model, how do I ensure model A is only trained on data set A and model B is only trained on data set B without deleting other mappings? In other words, I am seeing if there is a way that when I have all data set A and B labeled, I can select which labeled data I want to use to train each model.I want to do this because the formats of those documents can vary a bit, and I am really interested in the composed model feature as it has the ability to classify and seek for the better model.
Hope this explanation clarifies my question a bit better, thank you.
Hello @Mason , with custom extraction model you can label the document and train the models. However, it is not possible to select label data; you must either add your chosen labels and train the model, then reapply them to the same document and train the model.