Dataset and fine-tuned models from the study "Benchmarking and Adapting Compact Multimodal Models for Sinhala Visual Question Answering."