The mission

LeMaterial is an open-science initiative at the intersection of materials science and machine learning, created to accelerate the discovery of novel materials and deepen our understanding of the chemical space.

Our mission is to build and maintain a suite of harmonised datasets, benchmark tasks, machine learning models, and collaborative tools that empower researchers across disciplines, from quantum chemistry and experimentalists to ML scientists and software engineers, to drive forward an open and more effective scientific ecosystem.

Why LeMaterial?

The materials science ecosystem is evolving rapidly, but remains deeply fragmented. Datasets are scattered across platforms, often inconsistently formatted or missing key metadata. Benchmarks, models, and tools are siloed, making it difficult to build on others’ work or compare approaches fairly. LeMaterial aims to address this by creating a central, harmonized space where data, models, and ideas can move freely.

LeMaterial aspires to be more than just a repository, more like a connective tissue for the community, bridging computational chemists, ML researchers, software engineers, and experimentalists. We believe progress happens when people with different expertise align early around shared challenges, not when they work in isolation.

By pooling skills, sharing infrastructure, and co-designing tools and standards, we hope to enable more ambitious, interdisciplinary science that no single group could tackle alone. LeMaterial offers early-career researchers a chance to contribute meaningfully, while empowering senior scientists to help shape foundational resources for the field.

Above all, our goal is to accelerate discovery while lowering barriers to entry — so that any researcher with a good idea, regardless of background or affiliation, can help build the future of materials science

What You Can Work On

LeMaterial invites researchers from diverse domains to contribute to collaborative projects such as:

• Integrating new datasets (e.g. crystals, defects, surfaces, reactions) and expanding the range of available properties
• Developing predictive and generative ML models for materials and molecules
• Expanding analysis tools to explore chemical and structural landscapes
• Designing and maintaining evaluation benchmarks and leaderboards
• …and more!

🤝 An Open Invitation to Collaborate

LeMaterial is a platform for the community, built by the community. We welcome PhD students, postdocs, researchers, and professors to join forces on high-impact projects.

Our collaborative working groups are open to all, and all datasets, tools, and models are released under permissive open-science licenses.

While the initiative receives support from Entalpic and Hugging Face (for hosting, compute, and visibility), its direction is shaped by an open, distributed network of contributors.

📬 Ready to Contribute?

• 👉 Join the LeMaterial Slack
• 👉 Suggest your own idea for LeMaterial

Whether you’re contributing code, data, theory, or just curiosity — we would love to have you involved.

Let’s build the future of materials science — together.

Further reading

• Explore our blogpost for more details on the LeMaterial!