OpenAI

Playground for using OpenAI ChatGPT and Hugging Face state-of-the-art diffusion models for image, text and audio generation in PyTorch.

Local Installation to use the Hugging Face models

Download the Package and install the needed libraries with

python -m pip install --upgrade -r --force-reinstall requirements.txt

Create Python Environment

conda create --name myenv python=3.10
conda activate myenv

Run the software with

python -m streamlit run 🤖_OpenAI.py

Click on Hugging Face in the left sidebar and choose a model from the selectbox. The model will be automatically downloaded (stored in User/.cache folder). Create the file .streamlit/secrets.toml with this content:

[openai]
key = "openai-api-key"
image = "images/BenBox_small.png"

[hugging_face]
key = "hugging-face-key"

Llama.cpp

Get the sources, build and run the Llama3 model with the following commands:

 git clone https://github.com/ggerganov/llama.cpp.git
 cd llama.cpp
 make
 ./llama-server -m ~/.cache/lm-studio/models/DrBenjamin/llama-3-8b-chat-doctor/llama-3-8b-chat-doctor-Q4_K_M.gguf --port 1234 -c 2048

Fine-tuning the Llama3 model

Example of fine-tuning the Llama3 model on a medical conversation dataset.

Link to Kaggle Link to Wandb Link to Huggingface

Streamlit Cloud Demo

Snowflake Native App Framework

Python modules in Snowflake

brew tap snowflakedb/snowflake-cli
brew install snowflake-cli
python -m pip install --upgrade --force-reinstall snowflake-cli-labs

Snowflake Data Warehouse

The snowflake native app

Misc

Miscellaneous files and scripts for the project.

git repack -a -d -f --depth=250 --window=250
git gc --aggressive --prune