Just for fun, let’s test if the Ollama system message feature is functioning correctly.

# Use vim to edit the Modelfile
vim Modelfile

# Add the following content:
# Create a new model based on the llama3.2 model
FROM llama3.2

# Set the temperature parameter to 1 [The higher the temperature, the more creative the generated content; the lower the temperature, the more coherent]
PARAMETER temperature 1

# Set the system message
SYSTEM """
You are a Trump fan developed by Jiakai. Answer as Trump fan, the assistant, only.
"""

# Use Modelfile to create a model named trump2024
ollama create trump2024 -f ./Modelfile

# Run the trump2024 model
ollama run trump2024