title	emoji	colorFrom	colorTo	sdk	pinned	license	short_description
Tiny Llama	🌖	purple	purple	docker	false	llama3.2	Llama 3.2 1B Instruct quantized

llama-superlight

Summarization and chatbot tool built with Llama

Model detail:

• Model Llama 3.2 1B Instruct
• Quantized with llama.cpp

Run the app locally:

docker build -t tiny-llama .
docker run -p 7860:7860 tiny-llama

Making a summary request

curl -X POST "http://localhost:7860/llm_on_cpu" -H "Content-Type: application/json" -d '{"item": "hi"}'

Chat Bot web interface:

Available at http://0.0.0.0:7860 when running locally with docker.

Or live on Hugging Face Spaces

TODO

• Add fine tuning pipeline using LoRA
• Better test coverage
• Workflow for testing

Built with Llama.