Get Full Government Meeting Transcripts, Videos, & Alerts Forever!
Get email alerts on the Materials Discovery topic
No spam. Unsubscribe anytime.
DOE demo links hypothesis agents and MIST model to speed materials discovery
Summary
In a demo the presenter said a DOE agent proposes hypotheses and hands candidates to MIST, a University of Michigan materials foundation model, which scores properties and produces summaries and figures for inspection; the speaker requested "10 initial electrolyte candidates" as an example.
An Agency official for the Department of Energy demonstrated a workflow in which a DOE-developed hypothesis agent proposes experimental candidates and hands them to MIST, a materials foundation model from University of Michigan collaborators, to score properties and return results for reporting.
The presenter said, "Here, we ask for 10 initial electrolyte candidates for a next generation battery. Seconds later, we have a starting set with the reasoning laid out where we can inspect it." The demo highlighted automated hypothesis generation, scoring by MIST and automated report and figure generation as an end-to-end materials-discovery pipeline.
