-
Model : Llama 2 7b chat
-
Dataset : finance_questions by 4DR1455
-
Maximum Length : 256
-
Fine Tuning:
- QLoRA Adapter on Transformer Attentions
-
UI:
- Streamlit simple chat UI
Note:
If you already download the base pretrained model from huggingface you can create model_cache
folder inside this repository place the base pretrained model into model_cache
folder.
To deploy this project run
docker build -t finance-chatbot-app .
docker run -d --restart always --gpus all --name finance-chatbot-app -p 8501:8501 finance-chatbot-app
To stop and delete the container run
docker stop chatbot-app && docker rm finance-chatbot-app