The OpenLAM Challenges
This addresses the problem of limited evaluation datasets for researchers in materials science and computational chemistry, though it is incremental as it builds on existing LAM development efforts.
The paper tackles the lack of reliable benchmarks for evaluating Large Atom Models (LAMs) in scientific computation by establishing comprehensive benchmarks, resulting in the collection of over 19.8 million valid structures including 1 million on the OpenLAM convex hull to drive advancements in generative modeling and materials science.
Inspired by the success of Large Language Models (LLMs), the development of Large Atom Models (LAMs) has gained significant momentum in scientific computation. Since 2022, the Deep Potential team has been actively pretraining LAMs and launched the OpenLAM Initiative to develop an open-source foundation model spanning the periodic table. A core objective is establishing comprehensive benchmarks for reliable LAM evaluation, addressing limitations in existing datasets. As a first step, the LAM Crystal Philately competition has collected over 19.8 million valid structures, including 1 million on the OpenLAM convex hull, driving advancements in generative modeling and materials science applications.