Official Implementation of InstructZero; the first framework to optimize bad prompts of ChatGPT(API LLMs) and finally obtain good prompts!

High-quality instruction–response dataset for Kenyan financial reasoning, filtered via InstructZero-style validation using a LoRA-finetuned DistilGPT2 model. Each example is automatically checked for response consistency across multiple instruction rewrites, enabling more reliable instruction tuning.

This repo contains the kenyan_finance_selfinstruct_v6_instructzero.jsonl dataset — a zero-shot instruction dataset aligned with the InstructZero framework, adapted for the Kenyan personal finance domain.