This is a basic chainlit web app to run Llama2 in CPU machines. Its knowledge is constrained to the pdf file you store in the data folder. It is a simple way to get started with Llama2.
-
Clone the repo
-
Download the model from the link below and put it in the model folder at root level Llama2-7B-Quantized model
-
Install the requirements using pip:
pip install -r requirements.txt
-
Get your knowledge source into data folder and run the following command:
Depending on the size of your knowledge source, this might take a while.
python ingest.py
This will create a directory called vectorstores in the root folder. This is where the knowledge source is stored in a vectorized format.
-
Run the following command to start the web app:
chainlit run model.py -w
This will start the web app on port 8000. You can access it at http://localhost:8000
-
Ask questions and enjoy!