/LLM-Testing

Playground to test local LLMs with ollama, with functionalities such as function calling

Primary LanguageJupyter Notebook

LLM-Testing

Playground to test local Large Language Models with ollama, implementing functionalities such as function calling.

Setup

  1. Execute the following steps to install ollama's binaries and libraries, and create the models used in the playground.
    • Note: These steps correspond to Linux OS. For other OS, check ollama's webpage for installation instructions, then follow from step B.
# A. Install ollama (Linux)
curl -fsSL https://ollama.com/install.sh | sh

# B. Install required Python packages
pip install -r requirements.txt

# C. Run ollama server as a service
ollama serve

# D. Create models
ollama create weatherer --file weatherer/Modelfile  # Mistral 7B (4.1 GB disk, 5.8 GB memory)

  1. List installed models with ollama ls

  2. List mounted models with ollama ps

  3. Check the notebooks containing each tested functionality.

Uninstall

  1. Remove all custom models with ollama rm <Model Name>

  2. Follow ollama's instructions here (Linux).