R. Bao, “Time-Frequency Token Advantage Clipping for Training Efficient Large Reasoning Model”, AAAI, vol. 40, no. 36, pp. 30049–30057, Mar. 2026.