LLAMA-2 API + Python / JupyterLab

Source

• https://huggingface.co/docs/transformers/main_classes/text_generation

Setup Workspace Environments

1. register by meta to download the model

• https://ai.meta.com/resources/models-and-libraries/llama-downloads/

2. Create a Hugging Face Account

• site: https://huggingface.co/

3. Create a HF Token

• HuggieFace -> Settings -> Access Token > Create

5. install python packages

   pip install -r requirements.txt
   python -m pip install --upgrade transformers

6. Test the Token

   huggingface-cli login

7. Request Llama-2 model access

• site: https://huggingface.co
• model: meta-llama/Llama-2-7b-chat-hf
• action: Request Submit
• It may take a day to get access

8. Receive the permission notification by Email

Run Llama 2 in JupyterLab

1. install packages

   pip install -r requirements.txt

2. start jupyter lab

   jupyter lab 

3. Open Notebook

   cd notebook/step-1.ipynb

4. Login to HF to get access to the model

5. Execute scripts in Notebook

• step-1.ipynb
• step-2.ipynb

Run Llama 2 in your IDE

1. Login HuggingFace

   huggingface-cli login

2. install packages

   pip install -r requirements.txt

3. download model

   python src/main.py