This model is a BERT-based extractive question-answering model fine-tuned on the MechQA dataset, a domain-specific QA dataset focused on mechanical engineering materials and their properties. Achieves 78.03 EM / 84.50 F1 on the MechQA validation set.
This model was developed as part of the study: “Automatic Generation of a Mechanical Properties Question-Answering Dataset for Language Model Benchmarking: A Comparative Study of BERT, XLNet, and LLaMA Models”