This model is an XLNet-based extractive question–answering model fine-tuned on the SQuAD v1.1 dataset. It represents a general-domain QA baseline used to compare against domain-adapted models trained on the MechQA dataset. General-domain baseline model achieving 35.27 EM / 52.22 F1 on MechQA.
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”
