Scientific LLM Benchmarks
GitHub
← All benchmarks
Physics & Astro· physics-discovery

DiscoverPhysics

Princeton / NYU / Flatiron Institute / Polymathic AI · 2026

22 simulated worlds with non-standard physics where an agent proposes initial conditions, observes noisy N-body trajectories, and submits a natural-language law plus a Python implementation.

Physics
GitHub stars
Task type
agentic
Modality
text
Access
open
Size
22 items
License
MIT
Metrics
LLM-judge score, trajectory match

No sample rows available for this dataset.

Open it on Hugging Face ↗