PDECert Natural-Candidate Pilot Benchmark

License: See LICENSE file · Updated 2026-09-01
My rating

This dataset provides a pilot benchmark for checking symbolic candidate solutions to partial differential equations. It contains 20 records with unedited generator outputs, fully instantiated verification cases, content digests, producer metadata, and completed human annotations. The benchmark includes 10 symbolic-solver outputs and 10 open-model outputs, with equal numbers of valid and invalid cases.

Tags: