Towards Fully-Automated Materials Discovery via Large-Scale Synthesis Dataset and Expert-Level LLM-as-a-Judge
Heegyu Kim *, Taeyang Jeon *, Seungtaek Choi, Ji Hoon Hong, Dong Won Jeon, Ga-Yeon Baek, Gyeong-Won Kwak, Dong-Hee Lee, Jisu Bae, Chihoon Lee, Yunseo Kim, Seon-Jin Choi, Jin-Seong Park, Sung Beom Cho, Hyunsouk Cho · 2025
Read Towards Fully-Automated Materials Discovery via Large-Scale Synthesis Dataset and Expert-Level LLM-as-a-Judge by Heegyu Kim *, Taeyang Jeon *, Seungtaek Choi, Ji Hoon Hong, Dong Won Jeon, Ga-Yeon Baek, Gyeong-Won Kwak, Dong-Hee Lee, Jisu Bae, Chihoon Lee, Yunseo Kim, Seon-Jin Choi, Jin-Seong Park, Sung Beom Cho, Hyunsouk Cho (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-16457.txt.
Description
Heegyu Kim and co-authors present a computing paper from 2025 on how new materials are made. Making a material still rests on trial and error guided by expert gut feel. The team built a large data set to help language models learn the craft.
The set holds seventeen thousand recipes checked by experts, drawn from open papers. From it they built a test suite that covers the full task chain. The tasks run from picking raw goods and tools to writing steps and guessing the result. They also use a language model as a judge to score answers. Its scores agreed well with expert scores. The authors say the set and tests lay ground for fuller automation of materials work.
Catalog record
| CREATOR | Heegyu Kim *, Taeyang Jeon *, Seungtaek Choi, Ji Hoon Hong, Dong Won Jeon, Ga-Yeon Baek, Gyeong-Won Kwak, Dong-Hee Lee, Jisu Bae, Chihoon Lee, Yunseo Kim, Seon-Jin Choi, Jin-Seong Park, Sung Beom Cho, Hyunsouk Cho |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-16457