Skip to content

OpenWebUI

The OnDemand OpenWebUI app serves as a web interface for interacting with the Ollama platform.

Basic Configuration

To begin an OpenWebUI OnDemand session, the basic configuration we recommend is less than 4 hours (unless you require more), at least 4 cores, 4GB memory, and 1 GPU. To set the number of GPUs, you will first need to select the Use GPUs option.

OpenWebUI session config with 'Number of hours' set to 1, 'Number of cores' set to 4, 'Amount of memory' set to 4GB, using GPUs selected with 'Number of GPUs' set to 1

When you select to use GPUs, you also have the option to select GPU type. The GPU types you can select from are listed in a table below the GPU type dropdown.

OpenWebUI session config with a single h200 GPU

Advanced Configuration Options

Beyond these basic configuration parameters, there are more session configurations you can make by selecting the Advanced Options box below the set of basic configuration options. Here, you can specify the node type for the session to run on. Note that if you're using GPUs, the available GPU types available will depend on your selected node type.

OpenWebUI session config with node type set to amd24.

OpenWebUI session config with node type set to amd24 and GPU type set to h200.

As the OpenWebUI app supports tool plugins, and offers RAG capability, there are additional advanced options to configure your OpenWebUI session. This includes RAG Top K to set the number of results to consider when using RAG, Database URL to specify a database URL to connect to for a knowledge base, and Tools Directory for tools that customize and enhance the LLMs in OpenWebUI.

OpenWebUI session config with RAG and tools configurations.

Lastly, in order to launch the OpenWebUI app, you will need to read and agree to the terms of the MSU Interim Guidance on Data Uses and Risks of Generative AI.

OpenWebUI session config with the 'MSU Interim Guidance on Data Uses and Risks of Generative AI terms' confirmation box checked.