Rubric-Based Rewards for RL by Cameron R. Wolfe, Ph.D.

Extending the benefits of large-scale RL training to non-verifiable domains...