Hello.
I have a robot running using Document understanding (Modern) to extract data from PDF’s. Sometimes we make changes in the annotations or add more documents to the model in order to improve performance. My issue and question is, do i have to publish a new version of the model AND then open the robot and manually update the required values (Datetypes, model numbers for extraction activity) or is there a way were i can update my robot without having to manually changing values?
Hello @kts1!
It seems that you have trouble getting an answer to your question in the first 24 hours.
Let us give you a few hints and helpful links.
First, make sure you browsed through our Forum FAQ Beginner’s Guide. It will teach you what should be included in your topic.
You can check out some of our resources directly, see below:
Always search first. It is the best way to quickly find your answer. Check out the 
icon for that.
Clicking the options button will let you set more specific topic search filters, i.e. only the ones with a solution.
Topic that contains most common solutions with example project files can be found here.
Read our official documentation where you can find a lot of information and instructions about each of our products:
Watch the videos on our official YouTube channel for more visual tutorials.
Hopefully this will let you easily find the solution/information you need. Once you have it, we would be happy if you could share your findings here and mark it as a solution. This will help other users find it in the future.
Thank you for helping us build our UiPath Community!
Cheers from your friendly
Forum_Staff
Did you get the answer? I am on the same question, looking for the answer
In modern Document Understanding (DU), whenever you make any changes in annotations or annotate new documents, training will start automatically. Once training is complete, you need to publish the new version and also update it in your code, specifically in the extraction activity.
I don’t agree with this.
If you label new documents it won’t automatically trigger any re-training, you have to do that yourself and then once the training it done set the newly trained model as the ‘active’ one in the AI Center.
At that point you do not need to change your automations, it will be using whichever version you have set in the AI Center.
If you add new fields to a document thats a trickier job, not only do you need to retrain but also update code.
Hi @Jon_Smith,
The query is about is modern DU not AI Center,
Modern DU still deploys a model in the AI Center though right?
In Modern DU we have Active Learning which brings a new unified experience to training, deploying and monitoring your model in one user interface.
Let me double check, as they changed alot recently, to validate my assumptions are correct.
Apologies, you are correct, it seems the modern DU projects have been completely separated from the AI Center.
The automatic training seems abit annoying, but maybe I just need to get used to it.
I believe what I say stands about not needing to update your code however.
If you update the published version in the Document Understanding Center any automations using that project and extractor will automatically point to the new one.
You will need to update the project version manually in Studio. Since the Classifier and Extractor models are trained continuously as you upload data or interact with the model, you can use the Project version feature to freeze the current state of the models if you are satisfied with their performance.
What project version in Studio are you referring to? The Extract Document Activity in studio doesn’t refer to a specific version
You can use Document Understanding Project Extractor activity in data extraction scope in DU process template
Ah right.
I was referring to the Modern Cross Platform DU activities, not the Windows classic ones.
In the Modern you don’t specify the version.
Where in ‘Extract Document Data’ do you change the version if no version can be specified in that activity?
the ending _v3 _v4 changes and when we update to get new model we need to change and also accordngly the output type also would change..This has been reported as well.
cheers
Oh man, you’re right. I completely glanced over the version thing in the extractor name.
The last real DU project I did was before they migrated to the new experience in the DU Center, so when I was using the Modern activities it didnt need to be updated.
Thanks for pointing that out, thats important to know, and frustrating it messes with the generated data types.
