Deep Forest Sciences, in collaboration with the University of Michigan, received a DOE INCITE Award that provides access to 200,000 Polaris node hours to advance research on foundation models for inorganic materials.

This project, led by [Venkat Viswanathan](https://aero.engin.umich.edu/people/viswanathan-venkat/),  focuses on using large language models (LLMs) to accelerate the design of battery electrolytes. The effort builds on the [CLOUD](https://changwenxu98.github.io/talks/2024-06-19-Cloud) representation for crystalline materials led by Changwen Xu, as well as earlier innovations like ChemBERTa, introduced by Dr. Bharath Ramsundar. Supported by the [Michigan Institute for Computational Discovery & Engineering (MICDE)](https://micde.umich.edu/) and Argonne’s cutting-edge Polaris supercomputer, the team will train models on massive datasets, scaling up to 50 billion molecules to uncover new scientific insights. These advanced models aim to bridge the gap between computational predictions and experimental validation, enabling rapid testing of promising compounds.

To read more about the project, visit the [ASCR article](https://ascr-discovery.org/2024/10/ai-turbocharge/) featured by the U.S. Department of Energy Office of Science.

Deep Forest Sciences is a deep tech R&D company building Prithvi, an AI-powered scientific discovery engine for the biotech/pharma industries. Deep Forest Sciences leads the development of the open source DeepChem ecosystem. Partner with us to apply our foundational AI technologies to hard real-world problems in drug discovery. Get in touch with us at partnerships@deepforestsci.com!

Deep Forest Sciences Receives DOE INCITE Award with University of Michigan

Deep Forest Sciences has been selected as a finalist in the highly competitive [Standard Industries Chemical Innovation Challenge](https://www.herox.com/SIChallenge), sponsored by Standard Industries and [W. R. Grace & Co](https://grace.com/). The contest aims to transform small-molecule synthesis by developing AI/ML-based retrosynthesis tools guided by chemical intuition.

Retrosynthesis has been a major challenge in computational chemistry and AI, and has been a key research focus for CEO and founder of Deep Forest Sciences, Bharath Ramsundar, since his [PhD](https://pubs.acs.org/doi/full/10.1021/acscentsci.7b00303). The Prithvi™ retrosynthesis platform builds on Deep Forest Sciences expertise in chemical foundation models and computational chemistry and represents a new state-of-art for retrosynthesis. Deep Forest Sciences will deploy and test our breakthrough retrosynthetic technology with customers and partners over the coming year.

Deep Forest Sciences Selected as Finalist in Standard Industries Chemical Innovation Challenge

Deep Forest Sciences is excited to announce its collaboration with the University of Michigan to launch the [Scientific Foundation Models (SciFM)](https://scifm.ai/) initiative led by UM Associate Professor [Venkat Viswanathan](https://aero.engin.umich.edu/people/viswanathan-venkat/). Supported by a one-year grant from the Department of Energy, the initiative has received the DOE INCITE award for 200,000 node hours on Argonne National Laboratory Polaris to train the foundation model, and has been selected as part of Microsoft’s Accelerating Foundation Models Research Initiative.

The SciFM initiative aims to leverage large amounts of unlabelled data in scientific domains. To start, the effort will focus on developing scientific foundation models for Generative AI in materials design, PDE modeling, and autonomous scientific discovery. Deep Forest Sciences’ expertise in molecular foundation models, particularly in the field of drug discovery, has given us significant insight into scientific foundation model design and development. This expertise will be leveraged to accelerate battery design among other applications. The SciFM initiative highlights Deep Forest Sciences’ commitment to using scientific foundation models to accelerate scientific discovery for the common good.

To learn more about the SciFM initiative, visit [SCIFM website.](SciFMhttps://scifm.ai/) MICDE’s press can be found [here.](https://news.umich.edu/building-a-chemical-gpt-to-help-design-a-key-battery-component/)

Deep Forest Sciences is a deep tech R&D company building Chiron, an AI-powered scientific discovery engine for the biotech/pharma industries. Deep Forest Sciences leads the development of the open source DeepChem ecosystem. Partner with us to apply our foundational AI technologies to hard real-world problems in drug discovery. Get in touch with us at partnerships@deepforestsci.com!

News