Figure 3(b) visibly labels its two reward panels as MATH and Countdown and shows random-mask SPG suffering severe late-training degradation while SPG+IM remains stable; however, the surrounding Section 4.2 prose describes the affected tasks as GSM8K and MATH. The fixed source is internally inconsistent about whether the non-MATH collapse panel is Countdown or GSM8K, so task identity for that panel remains unresolved rather than being silently normalized.

Figure 3(b) visibly labels its two reward panels as MATH and Countdown and shows random-mask SPG suffering severe late-training degradation while SPG+IM remains stable; however, the surrounding Section 4.2 prose describes the affected tasks as GSM8K and MATH. The fixed source is internally inconsistent about whether the non-MATH collapse panel is Countdown or GSM8K, so task identity for that panel remains unresolved rather than being silently normalized. · CiteArk

Reported and observed measurements

Assessments (0)

Runs (0)