H. Lim, “GrayKD: Distilling Better Knowledge from Black-box LLM via Multi-rationale Injection”, AAAI, vol. 40, no. 38, pp. 31997–32005, Mar. 2026.