Datasets and aligned LLM checkpoints for training small open-source language models to perform knowledge-grounded material evaluation.