G. Chrysos, “Beyond Neuron-Level Sparsity: Achieving Faithful and Interpretable LLMs with Mixture of Decoders”, AAAI, vol. 40, no. 47, pp. 39811–39812, Mar. 2026.