mz516/llama-3.1-lora-MechQA

This model adapts LLaMA-3.1-8B-Instruct to the MechQA task using LoRA fine-tuning. The model is trained on the MechQA dataset, which contains automatically generated SQuAD-style question–answer pairs related to engineering materials and their mechanical properties. Achieves 80.48 EM / 86.25 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”