Learning TinyLlama

Just one of the things I'm learning. https://github.com/hchiam/learning

TinyLlama is an open-sourced GPT/LLM licensed for commercial use and is small enough to run on edge devices (quantizable to 637 MB).

Official repo: https://github.com/jzhang38/TinyLlama

Official demo: https://huggingface.co/spaces/TinyLlama/tinyllama-chat

• official demo Python code: https://huggingface.co/spaces/TinyLlama/tinyllama-chat/blob/main/app.py
• official demo requirements.txt: https://huggingface.co/spaces/TinyLlama/tinyllama-chat/blob/main/requirements.txt

https://www.youtube.com/watch?v=Weprd1wKoQ4

https://arxiv.org/abs/2401.02385

me testing it out

Demo on Google Colab: https://colab.research.google.com/drive/1p60kYImfZM2M_RyISR6hoK0xocVu2spX

• Jupyter notebook export: https://github.com/hchiam/learning-TinyLlama/blob/main/TinyLlama_test.ipynb
• Python export: https://github.com/hchiam/learning-TinyLlama/blob/main/tinyllama_test.py

TODO:

• make JS call Python code? https://github.com/hchiam/node-plus-python

other things you might find interesting

https://github.com/hchiam/learning-localGPT

https://github.com/hchiam/learning-gpt4all