Hi, everyone upon the last few weeks I have been using AI Document Intelligence to read a lot of invoices from a job sight that I work with I have been able to create a custom model, 2 custom models, composed models and even write some python code to allow me to do multiple documents(pdfs) at once. Now just because I have been able to use it does not mean I understand majority of things from it. I will jump right into my questions this way I hopefully can get some answers on this. Is there a way to create a python script to monitor how many monthly documents I have left before I reach my threshold of 500 documents For the storage account I noticed I have a HOT inferred storage container what exactly is this? Also what is this container exactly holding? Is it holding the custom models I have created? I am not able to store these models locally on my computer? I noticed there is an SHARE section on the custom model which release some long key. I have taken that key and tried to give it to a friend that I know so he can continue working on the project ( basically thought it would allow him to have his own copy of my custom trained model) but this wasn't the case as a matter of fact I had to begin giving permission and more things that I was unaware of so pretty much I stopped since I did not know to do it. My question is, is there a way to export my custom model and give it to someone else? Lets say I gave someone else my API key how can I keep track of how many times they used it on there account vs how many times I used it on my account? To be honest I have more questions but I feel like getting an answer on any of these things will help at least point me in the right direction. If anyone has any links that I can read on about any of these things I do have time to read and do not mind.

Few questions on Document Intelligence AI & Storage

KEVIN S 20

Hi, everyone upon the last few weeks I have been using AI Document Intelligence to read a lot of invoices from a job sight that I work with I have been able to create a custom model, 2 custom models, composed models and even write some python code to allow me to do multiple documents(pdfs) at once.

Now just because I have been able to use it does not mean I understand majority of things from it. I will jump right into my questions this way I hopefully can get some answers on this.

Is there a way to create a python script to monitor how many monthly documents I have left before I reach my threshold of 500 documents
For the storage account I noticed I have a HOT inferred storage container what exactly is this? Also what is this container exactly holding? Is it holding the custom models I have created? I am not able to store these models locally on my computer?
I noticed there is an SHARE section on the custom model which release some long key. I have taken that key and tried to give it to a friend that I know so he can continue working on the project ( basically thought it would allow him to have his own copy of my custom trained model) but this wasn't the case as a matter of fact I had to begin giving permission and more things that I was unaware of so pretty much I stopped since I did not know to do it. My question is, is there a way to export my custom model and give it to someone else?
Lets say I gave someone else my API key how can I keep track of how many times they used it on there account vs how many times I used it on my account?

To be honest I have more questions but I feel like getting an answer on any of these things will help at least point me in the right direction. If anyone has any links that I can read on about any of these things I do have time to read and do not mind.

Azar 19,400 Reputation points

2024-04-06T13:46:12.2766667+00:00

Hi there KEVIN S

Thats a good question and thanks for using QandA platform

lets start with the answering So to monitor the number of monthly documents processed by your AI Document Intelligence service, you can use the Azure Cognitive Services Management SDK for Python. This SDK allows you to interact with the Azure Cognitive Services APIs programmatically, including retrieving usage information. You can use the Get Usage API to fetch the current usage statistics for your service and then incorporate this functionality into a Python script to track your document usage over time.

The HOT Inferred Storage Container that is optimized for frequently accessed data with inferred access patterns. It is often used for storing data that is accessed frequently and requires low latency access. Regarding your custom models, they are stored in Azure Cognitive Services, rather than in a storage container. The container you mentioned may contain other metadata or configuration related to your models, but the actual model data is managed by the AI service itself.

Currently, I guess there isn't a direct way to export custom trained models from AI Document Intelligence for sharing with others. But, you can do this by providing your friend access to your Azure resources, specifically the Azure Cognitive Services instance where your custom models are deployed.

To track API key usage, you can use Azure's monitoring and logging. it allows you to track usage metrics, monitor API calls, and set up alerts for specific usage thresholds. By enabling logging and monitoring for your Cognitive Services instance.

I suggest you to go throw the azure documentations for more info. the links are below.

Azure SDK for Python documentation

Azure Storage documentation

Azure Blob Storage hot and cool access tiers

If this helps kindly accept the answer thanks much.
KEVIN S 20 Reputation points

2024-04-06T16:04:18.85+00:00

Hi so if I am getting this correctly your saying the model is just saved directly with azure cognitive services and not on any type of storage containers that I can see. This is basically kept separate from the storage container.

I have tried to give my friend access but it will also use up my api calls right and not his own? Like there no way to keep his calls seperate and my own seperate?

as far as the others go I appreciate it I will try to see if I can find a way to make api calls to see how many calls are remaining on my subscription