LLM-SRBench: 239 Scientific Equation Discovery Problems

239 challenging problems across four scientific domains designed to evaluate LLM-based scientific equation discovery methods. The benchmark was created by author nnheui and last updated on 2025-04-20. It includes a category, LSR-Transform, which transforms common physical models into less common mathematical representations to test reasoning beyond memorization.

Use Cases

• Benchmarking LLM performance on scientific equation discovery based on the 239 problems.
• Testing model reasoning beyond memorization based on the LSR-Transform category.
• Evaluating symbolic regression methods across four scientific domains mentioned in the description.

Strengths

• Contains 239 specifically designed problems to prevent trivial memorization.
• Covers four distinct scientific domains for broad evaluation.
• Includes the LSR-Transform category to test advanced reasoning.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.

Provenance

Source

huggingface

Collection Method

Introduced as a benchmark; specific data gathering method is not detailed.

Time Range

null

Freshness

Last updated 2025-04-20 01:07:26.

Geography

null