Preprint
Aug 2026
Privileged Likelihood Is Not Automatically Value: Three Checks for Token Credit in On-Policy Self-Distillation
The results suggest that dense credit assignment through distillation can be effective when its likelihood-based scores are empirically validated as meaningful proxies for outcome-relevant credit, when this alignment does not hold.
Xuan-Phi Nguyen, Shrey Pandit, Yiran Zhao et al.
· 1 citation