Follow these steps to set up the project:

1. Clone the repo into both your local and Newton account as there will be transferring of between local and Newton accounts.
2. Create a virtual environment using conda:

   conda create -n military_bt python=3.10

3. Navigate to the backend directory and install the required packages:

   cd backend
   pip install -r requirements.txt

4. Pull the CodeLLaMa model from Hugging Face into your local directory using the Hugging Face CLI:

   huggingface-cli download codellama/CodeLlama-7b-Instruct-hf

   • If you don't have the Hugging Face CLI installed, follow the instructions here to install and learn more.
   • Navigate to your cache located at /home/username/.cache/huggingface/hub and locate the CodeLLaMa model.
   • Copy the model into the models/codellama directory.
5. Set up your .env file with the following variables:

   PINECONE_API_KEY=
   
   REMOTE_HOST=newton.ist.ucf.edu
   REMOTE_USER=
   SSH_KEY_PATH=
   REMOTE_DIR= # create a directory in Newton for the files to copy over
   
   BASE_MODEL_PATH= # This should contain the path to the directory within the snapshots folder
   LORA_ADAPTER_PATH=codellama-bt-adapter

6. Run the shell script to submit the job to Newton:

   ./start_dev.sh