Fine-tune a model
Fine-tune a model in zCLOUD, the GPU cloud, by creating a fine-tuning task in the console from a model, a training set and a validation set. The steps use the form with Advanced Settings off; leave that switch off.
Before starting
The form asks for the following:
- A model (required), chosen in the **Model** picker. See [Models](https://docs.zettabyte.com/docs/zcloud/models-data-storage/models).
- A training set and a validation set, created first: open [**Datasets**](https://console.zettabyte.com/library-datasets/list), and see [Datasets](https://docs.zettabyte.com/docs/zcloud/models-data-storage/datasets) for the steps. Both are required.
For a task that needs a custom command or framework, create a **Training** task instead: see [Run a training job](https://docs.zettabyte.com/docs/zcloud/tasks/run-a-training-job).
Create the fine-tuning task
- Open [**Fine-Tuning**](https://console.zettabyte.com/task-finetune/index) from the console menu, in the **AI Infrastructure Rental** workspace (chosen in **Switch Platform**).
- Select **Create Fine-Tuning Task**.
- In **Name**, enter a name for the task.
- Select **Model**, choose a model in the **Select model type** picker, then select **Confirm**.
- Select **Training Set**, choose a dataset, then select **Confirm**.
- Select **Validation Set**, choose a dataset, then select **Confirm**.
- Open **Computing Config**, which holds **GPU Count(Card)**, **CPU(Core)** and **Memory(GiB)**.
- In **Storage**, select **+ Storage Volume** or **+ S3**. Each mount directory must be unique. See [Storage and PVCs](https://docs.zettabyte.com/docs/zcloud/models-data-storage/storage).
- Select **Review + Create**.
Next steps
- [Tasks overview](https://docs.zettabyte.com/docs/zcloud/tasks/training-and-tasks): to compare the task types.
- [How billing works](https://docs.zettabyte.com/docs/billing/how-billing-works): to understand the wallet and how charges work.