Ollama & Open WebUI User Guide at UiS

Hosted Endpoints

• Ollama API: https://ollama.ux.uis.no
  • Currently no authentication needed. But soon there will be authentication.
• Open WebUI: https://openwebui.ux.uis.no
  • Use your UiS employee ID and password to login to the chat window.

🚀 Accessing the Ollama API

🔧 Prerequisites

• Ensure the Ollama server is running by opening the status page
• If it is not working, then contact Jostein Hagen Lindhom jostein.lindhom@uis.no or Theodor Ivesdal theodor.ivesdal@uis.no
• Confirm that the desired model (e.g., llama3.3) is available by running this command:

curl https://ollama.ux.uis.no/api/tags

• If a model you need is missing, then contact Jostein to pull the model.

🧪 Example: Generate a Completion

Use curl to send a prompt to the model:

curl https://ollama.ux.uis.no/api/generate -d '{ 
    "model": "llama3.3", 
    "prompt": "What is the capital of Norway?", 
    "stream": false 
}'

Full documentation for using the API is here

🐍 Generate a Completion with Python

🔧 Prerequisites

• Python >= 3.8
• ollama PyPi package

Code Example for generating a single completion

python from ollama import Client

# Set up client to be used for generating completions
client = Client(host="https://ollama.ux.uis.no")

response = client.generate(
    model="llama3.3", 
    prompt="What is the capital of Norway?", 
    temperature=0.7, # optional: controls randomness 
    max_tokens=100 # optional: limits the length of the response 
) 
print(response["text"])