Evaluation of small-scale language models (1B–2B parameters) on financial question answering using the FinanceBench dataset with zero-shot, RAG, and fine-tuning methods. - View it on GitHub