AMD Developer Cloud: Redeem Credits and Launch LLaMA-Factory

This guide explains how to register for and sign in to the relevant AMD developer platforms, claim a compute voucher, redeem it for AMD Developer Cloud Credits, and launch the LLaMA-Factory Notebook to fine-tune a large language model.

We recommend completing these steps in a desktop browser. Page names and button labels may change slightly as the platforms are updated. Refer to the current interface if you notice any differences.

Step 1: Open the Campaign Page and Register or Sign In

You can access the page in either of the following ways:

Option A: Use the Campaign Link

Open the following link:

Go to the AMD Developer Program cloud compute redemption page

Option B: Start from the AMD AI Developer Website

Go to the AMD AI Developer Program. After signing in, open My Benefits ("我的权益"), find the cloud compute or compute voucher benefit, and click Redeem ("兑换"). Refer to the image below to locate the entry point.

When you open the page for the first time, you will be asked to register and sign in. Follow the on-screen instructions to create an account, verify your email address or mobile phone number, and sign in.

Figure 1: Campaign and benefits entry points in the AMD AI Developer Program

Step 2: Claim and Redeem the Compute Voucher

Follow the instructions on the campaign page to complete the process shown below:

Figure 2: Redeeming a compute voucher for AMD Developer Cloud Credits

After submitting the application in step 02, wait approximately five minutes. A button for viewing the cloud compute voucher will appear in the redemption history section. Click it to display the redemption link shown in step 03.

Step 3: Launch the LLaMA-Factory Notebook

After the Credits have been added to your account, complete the following steps on the AMD Developer Cloud platform:

1. Open Radeon Cloud Gallery.
2. Enter Llama factory in the search box.
3. Find the Llama factory workspace in the search results.
4. Click Launch in the lower-right corner of the card.

Launching a workspace consumes Credits. Check your account balance before launching it, and stop the instance promptly when it is not in use.

Figure 3: Search for Llama factory and click Launch

Step 4: Wait for the Workspace and Open the Notebook

After launch, the platform automatically creates and initializes the workspace. Wait until the progress reaches 100% and the message "Your workspace is ready" appears, then click Open Notebook.

If initialization takes a while, keep the page open and wait for it to finish. Do not click Launch repeatedly.

Figure 4: Click Open Notebook when the workspace is ready

Step 5: Run the LLaMA-Factory Notebook

After JupyterLab opens, you will see the LLM Fine-tuning with LLaMA Factory Notebook. Run the Notebook and fine-tune a model with LLaMA-Factory according to your requirements.

Figure 5: Open the LLaMA-Factory fine-tuning Notebook and run the cells in order

Step 6: Stop the Instance When You Are Finished

After running the Notebook or fine-tuning the model, promptly stop the cloud instance if you no longer need it:

1. Return to the AMD Developer Cloud platform.
2. Click your profile picture in the upper-right corner and open the Profile page.
3. In the Active Instance section, find the running Llama factory instance.
4. Click Destroy Instance and confirm the action by following the on-screen instructions.
5. Refresh the Profile page and verify that the instance is no longer shown as Active or Running.

Important: Closing JupyterLab, the Notebook, or the browser tab does not stop the instance, and it may continue consuming Credits. Always click Destroy Instance when you are finished.

Before destroying the instance, save or download any training results, model weights, and logs that you want to keep. Temporary data may not be recoverable after the instance is destroyed.

Figure 6: Click Destroy Instance on the Active Instance card