Fastweb/FastwebMIIA-7B · Guidance needed: creating a custom Modelfile for Ollama testing

Hi everyone,

I’m trying to use the FastwebMIIA‑7B model with Ollama for local testing. I performed the GGUF quantization myself, but I’m not sure how to structure a proper Modelfile and template for optimal results.

From the Hugging Face documentation, it seems possible to customize the prompt and parameters using:

FROM hf.co/{username}/{repo}:<quant>
TEMPLATE """
  {{ if .System }}<|system|>
  {{ .System }}<|end|>{{ end }}
  {{ if .Prompt }}<|user|>
  {{ .Prompt }}<|end|>{{ end }}
  <|assistant|>
  {{ .Response }}<|end|>
"""
SYSTEM """<your system prompt>"""
PARAMETER temperature 0.1
PARAMETER top_p 0.9
PARAMETER repeat_penalty 1.1

But I’m missing the precise details:

What’s the recommended TEMPLATE syntax for conversational use‑cases with this model?
How should I include stop sequences (e.g. </s>) correctly?
Are there additional PARAMETERS (like num_ctx, max_tokens) I should define?
Is there an example Modelfile template available for this specific model?

Any guidance, examples or best practices would be really appreciated.
Thanks in advance