Summary of Question Answering System Of Bridge Design Specification Based on Large Language Model, by Leye Zhang et al.
Question answering system of bridge design specification based on large language model
by Leye Zhang, Xiangxiang Tian, Hongjun Zhang
First submitted to arxiv on: 26 Aug 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The proposed question answering system uses a large language model to answer questions about bridge design specifications. Three implementation schemes are explored: fine-tuning a pre-trained BERT model, parameter-efficient fine-tuning, and building a self-contained language model from scratch. The system is trained on a dataset of self-built question-answer pairs using TensorFlow and Keras, with the goal of predicting start and end positions of answers in given bridge design specifications. Experimental results show that full fine-tuning achieves 100% accuracy across training, validation, and test datasets, allowing for accurate extraction of answers to user questions. While other schemes perform well on training data, their generalization abilities need improvement. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper creates a special computer system that can answer questions about building designs. The system uses a big language model like the ones used in search engines. It tries three different ways to make this work: using an already trained model, making small changes to the model, and starting from scratch. The system is tested on a set of questions and answers about building design specifications. The results show that one way works really well, but the others need some more improvement. |
Keywords
» Artificial intelligence » Bert » Fine tuning » Generalization » Language model » Large language model » Parameter efficient » Question answering
