> ## Documentation Index
> Fetch the complete documentation index at: https://hobbyist-e43fa225.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Customize a premium healthcare AI model by fine-tuning it in Microsoft Foundry

> Learn how to fine-tune a premium healthcare AI model in Microsoft Foundry.

Fine-tuning lets you adapt a premium healthcare AI model to your data and domain tasks. This article covers prerequisites and steps to create a fine-tuning job, with links to the core fine-tuning workflow in Microsoft Foundry.

<Info>
  The healthcare AI models marked (preview) in this article are currently in *limited preview*. These models are intended and provided as-is for research and model development exploration. The healthcare AI models are not designed or intended to be deployed in clinical settings as-is. They are not intended for use in the diagnosis or treatment of any health or medical condition, and the individual models' performances for such purposes have not been established.

  You bear sole responsibility and liability for any use of the healthcare AI models, including verification of outputs and incorporation into any product or service intended for a medical purpose or to inform clinical decision-making, compliance with applicable healthcare laws and regulations, and obtaining any necessary clearances or approvals.
</Info>

## Prerequisites

* An Azure subscription. If you don't have one, [create one for free](https://azure.microsoft.com/pricing/purchase-options/azure-account?cid=msft_learn).
* A Foundry project in a region that supports your target fine-tuning model. If you don't have one, [create a project](../create-projects).
* Access to a [premium healthcare model](/models/healthcare-ai-models) such as [MedImageInsight Premium (preview)](/models/deploy-medimageinsight-premium) or [CxrReportGen Premium (preview)](/models/deploy-cxrreportgen-premium) in the model catalog for your project.
* The **Foundry Owner** role, or a custom role that includes permissions to fine-tune and deploy models.
* Training and validation files in JSONL format, encoded as UTF-8 with BOM, and less than 512 MB per file.
* A healthcare data governance process that covers de-identification, retention, auditing, and access control.
* Review the core guidance before you train:
  * [Customize a model with fine-tuning](/models/fine-tuning)
  * [Microsoft Foundry fine-tuning considerations](/models/fine-tuning-considerations)
  * [Quotas and limits for Azure OpenAI in Foundry](/models/quotas-limits)

## Fine-tune a premium healthcare AI model

To create a fine-tuning job in the Foundry portal:

1. Sign in to [Microsoft Foundry](https://ai.azure.com/?cid=learnDocs). Make sure the **New Foundry** toggle is on. These steps refer to **Foundry (new)**.

<img src="https://mintcdn.com/hobbyist-e43fa225/_qpHdwibkfCcXaky/images/new-foundry.png?fit=max&auto=format&n=_qpHdwibkfCcXaky&q=85&s=1338a0cf43c92807e8bcccdd0223d052" width="184" height="36" data-path="images/new-foundry.png" />

1. Select your subscription and Foundry resource.
2. Go to **Build** > **Fine-tune**, and select **Start fine-tuning**.
3. Select the premium healthcare model you want to customize. You can also select a previously fine-tuned model.
4. Choose the customization method that your selected model supports: **Supervised** for supervised fine-tuning (SFT), or **Direct Preference Optimization** (DPO), or **Reinforcement** for reinforcement fine-tuning (RFT).
5. Select the training type: **Standard** (in-region, data residency), or **Global** (lower cost, faster queue), or **Developer** (experimentation, preemptible).
6. Upload or select your training and validation datasets.
7. Optionally, configure a suffix, seed, and hyperparameters.
8. Select **Submit** to start the job.

After the job completes:

1. Review training metrics and checkpoints on the job details page.
2. Confirm safety evaluation status before deployment.
3. Select **Deploy** on the job details page to deploy the fine-tuned model.
4. Test with a representative validation set and keep qualified human reviewers in the workflow.

For the full fine-tuning workflow, including data preparation, format requirements, hyperparameter options, monitoring, and deployment, see [Customize a model with fine-tuning](/models/fine-tuning).

For model-specific fine-tuning details, including supported methods and training parameters, see the model card for your chosen model in the Foundry model catalog.

## Related content

* [Healthcare AI examples (GitHub)](https://aka.ms/HealthcareAIExamples)
* [Customize a model with fine-tuning](/models/fine-tuning)
* [Deploy a fine-tuned model](/models/fine-tuning-deploy)
* [Fine-tuning safety evaluation](/models/fine-tuning-safety-evaluation)
* [CxrReportGen Premium healthcare AI model](/models/deploy-cxrreportgen-premium)
* [MedImageInsight Premium healthcare AI model](/models/deploy-medimageinsight-premium)