Can you share the full deploy code

#1
by tomriddle - opened

Can you share the full deployment code which you have deployed on other server and using end point to confirm that this space is miqu

  1. Download ollama from github
  2. Create Modelfile card
# Modelfile for a Q&A model with creative responses
FROM ./miqu.gguf

# Setting the temperature to 1 for more creative responses
PARAMETER temperature 1
PARAMETER top_p 0.95

# Template for processing inputs in the format of <s> [INST] QUERY_1 [/INST] ANSWER_1</s> [INST] QUERY_2 [/INST] ANSWER_2</s>...
TEMPLATE """{{- if .First -}}<s>{{- end -}}[INST] {{ .Prompt }} [/INST] {{ .Response }}</s> """
  1. Download Miqu.gguf with weights of that model (You can find link here )

  2. Start ollama serving (ollama serve)

  3. Replace LLM service url to your own server IP

  4. Enjoy!

Sign up or log in to comment