Verifiable env

Hit the metric. Any solution counts.

• —Same format as the other RL envs, but metrics replace the judge — no researcher taste in the reward, so diverse solutions all score.
• —If a problem returns a flat 0, the prompt is revised to give hints.
• —The value is the set of hard problems (exactly like MLS-Bench) plus researcher intermediate steps to de-sparsify them.

Samples

• Analogical RL Representation Comparison—
• Augmentor Policy Handoffs—
• Can a decoding method that measures novelty in a …—
• CRL Importance Sampling—
• How close to optimality are current state-of-the-…—
• Is there a characteristic geometric structure tha…—
• Long-Horizon Offline GCRL Value Errors—
• RL For Novel LLM Capabilities—