Chippy: A minimal prompting interface for Chip

Chippy provides a minimal prompting interface for chip_1.4B_instruct_alpha, a new experimental instruction tuned model based on pythia-1.4B-deduped. With Chippy, you can easily prompt user input via Streamlit and FastAPI. Please keep in mind that quick basic prompting is the ONLY purpose of this demo.

For a more serious project, check out OpenAssistant:

https://github.com/LAION-AI/Open-Assistant

Run in Google Colab

If you don't own a >=8GB GPU, you can use Google Colab by importing the notebook from /colab. It's the Streamlit app without the FastAPI.

If you want to run the app and the API locally, follow the instructions below.

Installation

To install Chippy's dependencies, you can use pip. Simply run the following command in your terminal:

pip install -r requirements.txt

Starting FastAPI

python api.py --host "127.0.0.1" --port 8000

Launching Streamlit

python app.py

Limitations of this demo app:

• Only single turn, no conversations yet
• This also means that the model won't remember previous prompts
• GPU only, minimum 8GB (CPU would be possible, but I found it to be way too slow)
• Hardcoded generation settings
• No support for the bigger versions of chip since this repo is targeted towards consumer hardware

For more information, check out the model card on Huggingface: https://huggingface.co/Rallio67/chip_1.4B_instruct_alpha

Thanks to LAION contributors and Stability.ai.