jill-stingray-sst05.py-script

Uploaded model

• Developed by: kittyjosh111

This is a cli script meant to interface with the Jill Stingray llama3.2 models I have on HuggingFace. It is based on my previous llm-interfacing scripts, which are on my GitHub. The repo for all the scripts is gptChat, and the subfolder for this script in particular is second-gen.

This is my first attempt at using tools with ollama. Honestly, I don't think I'm going to use this again unless I have a really precise use-case. Here, the tools are to make drinks or search about information on the drinks (in the calicomp csv), but the llm in charge of tools tends to be finnicky and work sporadically. I personally used qwen3:0.6b to handle tools, as its small yet good enough.

The model I trained does not seem to support tool roles. I'm not sure why, as the base model does. Anyway, I work around this by injecting tool output as a system role into the messages list, right before the user query.

Links

• VA11_HallA_Cyberpunk_Bartender_Action Steam

• Model: https://huggingface.co/chuanli11/Llama-3.2-3B-Instruct-uncensored

• Dataset: https://github.com/NoPlagiarism/va11halla-dialogues (did some formatting to make it a ShareGPT format)

The merged model can be found here

The gguf versions can be found here

<img src="https://raw.githubusercontent.com/unslothai/unsloth/main/images/unsloth%20made%20with%20love.png" width="200"/>