A medical chatbot that can talk about medicines, diseases, symptoms and treatments.
- Can chat about diseases and their symtopms, drugs and treatments etc.
- Utilizes RAG (Retrieval-Augmented Generation) with a medicine encyclopedia provided as a pdf. (The document can be changed to make the bot more intelligent)
- Uses HuggingFace embeddings for vector conversion.
- Uses FAISS(Facebook AI Similarity Search) to create and store an index on the local machine.
- Uses the llama 2 LLM model for generating natural response.
git clone https://github.com/vishnusingh-12/doc-bot cd doc-bot pip install -r requirements.txt python embedding_creation.py chainlit run model.py
The model uses GPU with CUDA 11.8 to run for which LlamaCpp has to be configured differently. Follow the below link for reference: https://python.langchain.com/docs/integrations/llms/llamacpp
The model uses a quantized llama 2 model and can also run on cpu with no change in the code. If LlamaCpp Cuda version is available it uses GPU else it uses CPU.
For support, contact me:
