Modelfile for a consistent local persona
A Modelfile that fixes the system prompt and sampling parameters of a local model, so the same behaviour holds in every session and you stop retyping instructions into each chat.
AI Place Hub
44
FROM llama3.3:70b
PARAMETER num_ctx 32768
PARAMETER temperature 0.3
PARAMETER top_p 0.9
SYSTEM """
You answer directly, without restating the question first.
When you are not sure something is correct - a function name, a version number, a fact you cannot verify from what was given - say so in one clause rather than presenting it with the same confidence as something you are sure of.
Keep answers proportional to the question. A short question gets a short answer.
"""
# Build with: ollama create mine -f Modelfile
# Run with: ollama run mineNotes from the author
The temperature and context values are a sensible default for instruction-following, not a universal answer. Raise the temperature for creative work. Raise num_ctx if you often paste long documents, because most Modelfiles default below what the model supports.
For information only, not professional advice. You use it at your own risk and we are not liable for any loss or damage. See the disclaimer.
Tagged