This project uses RAG to interact with CSV data. It allows users to ask natural language questions about the contents of a CSV file, and the system generates SQL queries to extract the relevant information, with integrated Google Generative AI and combining document loaders, embeddings, and vector stores for efficient data retrieval.
- RAG Workflow:
Learned to integrate a Retrieval-Augmented Generation pipeline effectively by combining embeddings with LLM reasoning.
- Embedding Management:
Understood the importance of efficient storage and retrieval of embeddings (FAISS).
- Query Interpretation:
Gained insights into parsing natural language queries and mapping them to structured csv data.
- LLM Integration:
Realized the potential of LLMs to enhance data retrieval and provide responses.
This assignment provided practical experience in building intelligent systems that combine machine learning and natural language processing for data mining tasks.
Step 1: Install Jupyter Notebook
Step 2: Get an API key of Google Gemini
Step 3: Install the dataset provided with this project
Step 4: install the libraries listed in the requirements.txt file which is present in the folder
Step 5: Run the ragCSV.ipynb file in Jupyter Notebook
Step 6: Inside the ragCSV file, enter the input in the line:
answer = rag_chain.invoke({"input": "enter the desired input here"})
Disclaimer: To run any separate csv file with this model, one needs to modify the system prompt slightly to match the dataset.