DevOps with Custom Model

Question

DevOps with Custom Model

Abhilash Kozhikkot (C) 20

Hello,

We have a custom form which we want to extract data from. We see the value of using a custom model for this purpose. However from the documentation its not very clear to me lets say we build a custom model, how do we then deploy this to other environments . e.g. UAT/ Production etc.

Can we still align with DevOps pipelines to do this ? Any information is highly welcome

Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-04T23:54:04.7666667+00:00

Hi Abhilash Kozhikkot (C),

Greetings & Welcome to the Microsoft Q&A forum! Thank you for sharing your query.

I understand that you are facing an issue with deploying a custom model to different environments such as UAT and Production.

To address your concern, you can align DevOps pipelines to automate this deployment effectively. By passing the target environment (env) as a parameter during pipeline execution, you can conditionally run tasks tailored for each environment. For instance, the pipeline can dynamically retrieve environment-specific configurations (e.g., endpoints, keys) from secure stores like Azure Key Vault and execute deployment commands based on the specified env. After training your model, export and register it in a model registry to maintain version control and ensure traceability. This approach enables consistent, scalable, and traceable deployments while supporting robust DevOps practices across multiple environments.

I hope this information helps. Thank you!
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-06T00:53:25.5133333+00:00

Hi Abhilash Kozhikkot (C),
Hope you are doing well.
We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution, please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Abhilash Kozhikkot (C) 20 Reputation points

2024-12-06T01:36:44.5533333+00:00

Hello The question is not about the devops pipeline .

Lets say I train a custom model with documents and I label fields and publish it and test with this. Once I am happy, how do I take this custom model to production. Do I need to do the labelling again on the production document intelligence instance also ?
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-06T17:09:32.1933333+00:00

Hi Abhilash Kozhikkot (C)
Hope you are doing well.
Thank you for clarifying!
No, you do not need to repeat the labeling process on the production Document Intelligence instance. Once you train, label, and publish a custom model, it is stored in the Document Intelligence resource where it was created.
To move the model to production, simply retrieve the model ID from your development environment and use it in your production environment. When making API calls in production, specify the production instance's endpoint and API key but reference the model ID from the trained model. This ensures the same model is used without the need for re-labeling, enabling a seamless transition to production while maintaining consistency.
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-09T17:08:58.1066667+00:00

Hi Abhilash Kozhikkot (C),
Hope you are doing well.
We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution, please do share that same with the community as it can be helpful to others.
Thank you!

Your answer

Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-04T23:54:04.7666667+00:00

Hi Abhilash Kozhikkot (C),

Greetings & Welcome to the Microsoft Q&A forum! Thank you for sharing your query.

I understand that you are facing an issue with deploying a custom model to different environments such as UAT and Production.

To address your concern, you can align DevOps pipelines to automate this deployment effectively. By passing the target environment (env) as a parameter during pipeline execution, you can conditionally run tasks tailored for each environment. For instance, the pipeline can dynamically retrieve environment-specific configurations (e.g., endpoints, keys) from secure stores like Azure Key Vault and execute deployment commands based on the specified env. After training your model, export and register it in a model registry to maintain version control and ensure traceability. This approach enables consistent, scalable, and traceable deployments while supporting robust DevOps practices across multiple environments.

I hope this information helps. Thank you!
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-06T00:53:25.5133333+00:00

Hi Abhilash Kozhikkot (C),
Hope you are doing well.
We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution, please do share that same with the community as it can be helpful to others. Otherwise, will respond with more details and we will try to help.
Abhilash Kozhikkot (C) 20 Reputation points

2024-12-06T01:36:44.5533333+00:00

Hello The question is not about the devops pipeline .

Lets say I train a custom model with documents and I label fields and publish it and test with this. Once I am happy, how do I take this custom model to production. Do I need to do the labelling again on the production document intelligence instance also ?
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-06T17:09:32.1933333+00:00

Hi Abhilash Kozhikkot (C)
Hope you are doing well.
Thank you for clarifying!
No, you do not need to repeat the labeling process on the production Document Intelligence instance. Once you train, label, and publish a custom model, it is stored in the Document Intelligence resource where it was created.
To move the model to production, simply retrieve the model ID from your development environment and use it in your production environment. When making API calls in production, specify the production instance's endpoint and API key but reference the model ID from the trained model. This ensures the same model is used without the need for re-labeling, enabling a seamless transition to production while maintaining consistency.
Pavankumar Purilla 4,980 Reputation points Microsoft External Staff

2024-12-09T17:08:58.1066667+00:00

Hi Abhilash Kozhikkot (C),
Hope you are doing well.
We haven’t heard from you on the last response and was just checking back to see if you have a resolution yet. In case if you have any resolution, please do share that same with the community as it can be helpful to others.
Thank you!

Share via

DevOps with Custom Model

Your answer