principia-ai / PhysGym
A benchmark suite for evaluating LLM-based interactive scientific reasoning.
Why this job: says: benchmark; topics: benchmark
What it does
- Controlled Prior Knowledge (L1-L4). Systematically vary what contextual information agents receive, from full descriptions to completely anonymized variables
- Interactive Physics Environments. 97 curated physics problems from PHYBench, each with executable simulations
- Realistic Constraints. Limited experimental budgets (100 experiments) that mirror real scientific practice
- Complete History Tracking. Monitor exploration dynamics, hypothesis evolution, and reasoning strategies
Get started
No single install command. Read the steps in the README (opens in new tab)
- Maintenance
- Inactive
- Last commit 10 months ago
- Stars
- 103
- +1 this week
- Licence
- MIT
- Allows business use
- Downloads for
- None listed
- No release files
- Latest version
- None yet
- No releases
- Built with
- Python
- 88% of the code
Is it looked after?
- Contributors
- Not stated
- Last 90 days
- Issues
- None
- No issues filed
- Pull requests
- 2 merged
- 1 waiting
- Releases
- None
- Never published
- No contributing guide
- No security policy
- No code of conduct
Similar repos
| Repo | Stars | Last commit | Licence | Downloads for |
|---|---|---|---|---|
| materialsproject / matbench | 225 | 2 days ago | MIT | No downloads |
| mlcommons / training | 1,778 | 16 days ago | Apache-2.0 | No downloads |
| alibaba / sec-code-bench | 137 | 31 days ago | Apache-2.0 | No downloads |
More: languages, topics and project record
Languages
- Python 88%
- Jupyter Notebook 10%
- Shell 2%
Topics
- agentic-ai
- ai-scientist
- benchmark
- equation-discovery
- llm
- physics
Project record
- JobEvals and observability
- StatusListed
- Commits5
- Forks14
- Watching9
- Stars in 28 days+2
- CreatedOct 22, 2025
- First seen by usNov 7, 2026
- CheckedNov 8, 2026