specialised assistant to talk with PDF

I have built a bot, so basically it answer the queries of the customer from the basis of the data in pdf but the issue is the replies are not accurate enough they are vague and it looks like the assistant is not smart enough to provide the relevant information, understand the query of the customer

Technologies Used

Python Libraries:

1. Streamlit: For building the web interface.
2. pdfplumber: For extracting text from PDF files.
3. scikit-learn: For TF-IDF vectorization and cosine similarity.
4. transformers: For integrating a free Hugging Face language model (flan-t5-small).

Other Tools:

1. Hugging Face Models: For text generation.
2. PyTorch: Backend for running the Hugging Face model.