Skip to content

Dataset · FineEnvs

Math problems in Harbor environments

FineEnvs/openai-math

369 Harbor environments with math problems from openai/math and verified Lean proofs.

What's inside

An experimental set of 369 Harbor environments with math problems ported from the openai/math release (commit adc7f12). In the release, 405 results are formalized as Lean statements; the set includes environments for which the official proof is verified from start to finish. The authors warn that the release is new and has not undergone independent verification, so rough edges are possible.

What you can do with it

  • Evaluate models and agents on mathematical proofs
  • Assess work with Lean formalization
  • Test Harbor environments for math
Tasks
Other
Size
up to 1K rows
Downloads this month
15.8K
Likes
135

Models trained on it

How to load it

from datasets import load_dataset

ds = load_dataset("FineEnvs/openai-math")