Skip to content

Harbor catalog

reasoning-gym/reasoning-gym-hard

Training/development/subset candidate

About reasoning-gym/reasoning-gym-hard

What it measures

Consult the upstream task verifier.

Scope & limitations

Name/card indicates a training, generated, development or subset collection; inspect held-out protocol before using it as an independent evaluation.

Evaluation environment

Dataset-specific harness qualification required.

Source

Original benchmark

Dataset & task catalog ↗