RAG_chatbot
"Bioinformatics chatbot using RAG for smart Q&A."
Created on 8th June 2025
•
RAG_chatbot
"Bioinformatics chatbot using RAG for smart Q&A."
The problem RAG_chatbot solves
Traditional LLMs, which are trained on static data, tend to hallucinate when asked questions about any specific domain, like bioinformatics, as they are not aware of recently developed tools and techniques.
To solve this problem, I have used RAG(Retreived Augmented Generation), which gives updated data to the LLM, which in turn gives a well-updated answer.
people can use it for bioinformatics study, research, protocol understanding.
Challenges I ran into
-There were a few challenges, like embedding of data, storage of embedded data, and a good well well-constructed chat-like answer for the question.
-For embedding, I have used the Huggingface model(sentence-transformers/all-mpnet-base-v2), which is efficient. I have used Hugging Face for both the embedding of the prompt as well as the data given to train.
-For storage, I had to go with Chromadb as it stores the vectorized data(embeds) locally, which makes the experience smooth and fast.
-For a good well well-constructed answer, I used meta-llama/Llama-3.2-3B-Instruct. this model gives a chat-like, well-presentable answer.
-My main focus in this project was the integration of RAG with LLM, so UI layer might not be that appealing.