Updating
Upgrading your RHEL AI system and models
Abstract
Chapter 1. Updating Red Hat Enterprise Linux AI
Red Hat Enterprise Linux AI allows you to update your instance so you can use the latest version of RHEL AI and InstructLab.
1.1. Updating your RHEL AI instance
You can update your instance to use the latest version of RHEL AI and InstructLab
Prerequisites
- You installed and deployed a Red Hat Enterprise Linux AI instance on one of the supported platforms.
- You created a Red Hat registry account.
- You have root user access on your machine.
Procedure
Log into your Red Hat registry account with the
podman
command:$ sudo podman login registry.redhat.io --username <user-name> --password <user-password> --authfile /etc/ostree/auth.json
Or you can log in with the
skopeo
command:$ sudo skopeo login registry.redhat.io --username <user-name> --password <user-password> --authfile /etc/ostree/auth.json
Upgrading to a minor version of Red Hat Enterprise Linux AI
You can upgrade your instance to use the latest version of Red Hat Enterprise Linux AI by using the
bootc
image for your machines hardware.$ sudo bootc switch registry.redhat.io/rhelai1/bootc-<hardware-vendor>-rhel9:<rhel-ai-version>
where
- <hardware-vendor>
-
Specify the hardare vendor for your accelerators. Valid values include:
nvidia
,amd
, andintel
. - <rhel-ai-version>
Specify the RHEL AI version you want to upgrade to.
Example command for machines with NVIDIA accelerators
$ sudo bootc switch registry.redhat.io/rhelai1/bootc-nvidia-rhel9:1.4
Restart your system by running the following command:
$ sudo reboot -n
ImportantWhen upgrading to RHEL AI version 1.4, the
/etc/skel/.config/containers/storage.conf
storage configuration needs to be copied to your<user-home>/.config/containers/storage.conf
home directory before re-initializing.For example:
$ cp /etc/skel/.config/containers/storage.conf <user-home>/.config/containers/storage.con
Upgrading to a z-stream version of Red Hat Enterprise Linux AI
If a z-stream exists, you can upgrade your system to a z-stream version of RHEL AI by running the folllowing command:
$ sudo bootc upgrade --apply
It is recommended to re-initialize your environment and configurations after you upgrade to the latest RHEL AI version.
WarningThe
ilab config init
command overrides your existingconfig.yaml
file and sets it to the default configurations per your system hardware. If you customized theconfig.yaml
file, ensure you are familiar with your custom configurations before re-initializing.$ ilab config init
Chapter 2. Updating a model
Red Hat Enterprise Linux AI allows you to upgrade LLMs that you locally downloaded to the latest version of the model.
Feature | Description | Version introduced | NVIDIA | AMD |
---|---|---|---|---|
| Granite student model for customizing and fine-tuning | 1.3 | Generally available | Technology preview |
| Granite model for inference serving | 1.3 | Generally available | Technology preview |
| Preview of the version 2 8b Granite model for inference serving | 1.3 | Technology preview | Technology preview |
| Version 1 of the Granite 3.1 base model for customizing and fine-tuning | 1.4 | Generally available | Technology preview |
| Version 1 of the Granite 3.1 model for inference serving | 1.4 | Generally available | Technology preview |
2.1. Updating the models
You can upgrade your local models to the latest version of the model using the RHEL AI tool set.
Prerequisites
- You installed the InstructLab tools with the bootable container image.
-
You initialized InstructLab and can use the
ilab
CLI. - You downloaded LLMs on Red Hat Enterprise Linux AI.
- You created a Red Hat registry account and logged in on your machine.
Procedure
You can upgrade any model by running the following command.
$ ilab model download --repository <repository_and_model> --release latest
where:
- <repository_and_model>
-
Specifies the repository location of the model as well as the model. You can access the models from the
registry.redhat.io/rhelai1/
repository. - <release>
-
Specifies the version of the model. Set to
latest
, or a specific version of the model, for the most up to date version of the model.
Verification
You can view all the downloaded models on your system with the following command:
$ ilab model list