This repository contains Jupyter notebooks for generating questions, answers, and context-based questions and answers using the Stanford Question Answering Dataset (SQuAD). The notebooks demonstrate how to work with the SQuAD dataset and generate natural language questions and answers.
-
Question Generation: This notebook focuses on generating questions from given text passages. It uses the SQuAD dataset to fine-tune a model for question generation.
-
Answer Generation: In this notebook, you can generate answers to questions based on a given context. It uses pre-trained models to identify answers within the context.
-
Context-Based QA: The context-based question and answer generation notebook combines the previous two functionalities. Given a context, it generates questions and answers. This is particularly useful for tasks like question-answering chatbots.
-
Chain Of Thought QLoRA based QA : Mistral 7B non instruct model finetuned on the SQUAD Dataset with comparable results. Trained till 9000 steps. (Link)[https://huggingface.co/vpgits/Mistral-7B-v0.1-qagen-v0.3]