PakLegalAid: Lawyer-Written Legal Question-Answer Dataset for Pakistani Law

A dataset of real legal questions answered by practicing lawyers, created for the PakLegalAid project. Built as part of the paper:"Enhancing Legal Assistance with Large Language Models: A Parameter-Efficient Fine-Tuning and Retrieval-Augmented Generation Approach" — Ahmed et al., submitted to PLOS ONE, 2025. Background:After fine-tuning LLaMA 3.2 on a legal corpus, the model produced academic, textbook-style answers. To address this, practicing lawyers wrote the responses directly — making model outputs sound like an actual legal professional rather than a law textbook. Dataset Structure:- Query: Legal question from a real user- Response: Answer written by a practicing lawyer Coverage includes family law, property law, tenancy, criminal law, immigration, corporate registration, and more — primarily within Pakistani legal context. This dataset was used for:1. Fine-tuning generative LLMs (LLaMA 3.x, Phi-3.5 Mini)2. RAG pipeline retrieval via FAISS vector store3. Evaluation of generated legal responses For research and fine-tuning only. Not a substitute for professional legal advice.