Heuristic development in the use of large language models for materials science
In the ever-expanding field of Artificial Intelligence (AI) tools, Generative Pre-trained Transformer (GPT) is one of many Large Language Models (LLMs) that has revolutionized how academia can interact with data. Simple sentences can be fed into GPTs to return long paragraphs of explanation, informa...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2024
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/176081 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | In the ever-expanding field of Artificial Intelligence (AI) tools, Generative Pre-trained Transformer (GPT) is one of many Large Language Models (LLMs) that has revolutionized how academia can interact with data. Simple sentences can be fed into GPTs to return long paragraphs of explanation, informative search results and even form discussions. This brought a spike in how information is utilised and shifted the entire tech sector in ways comparable to the birth of the internet. This feasibility study, we will harness GPT’s learning capabilities in the form of a chatbot that can bridge knowledge gaps within the field of Material Science. The method proposed is to enrich GPTs with context, through papers selected by a set of specialised search algorithm. By distilling essential research papers and textbook resources, the chatbot is tailored to generate detailed conversations within the user’s specified domain. The experimentations have used AI tools from the following: open-source models from Hugging Face, free literature database tool from Litmaps and paid model of GPT-4 by OpenAI. |
|---|