CuisineRAG is a modular Retrieval-Augmented Generation (RAG) system designed to answer questions about South Asian cuisine using a curated knowledge base. The system retrieves relevant contextual information from a dataset and uses a Large Language Model (LLM) to generate grounded responses.
- Open the main_notebook.ipynb , run all the cells one by one.
- Place your input file (in json format) in the "inputs_and_outputs" folder.
- You also have the option to change your output file location.
- After executing the run_json_input_output cell, you can view your output from this location.
- For evaluation on our benchmark dataset (which is named latest_benchmark.json and is placed in "data" folder) run the cell containing evaluate_rag_pipeline function.
- The evaluation metrics along with the scores are printed in the output of this cell.