Beyond Neuron-Level Sparsity: Achieving Faithful and Interpretable LLMs with Mixture of Decoders.

Grigorios Chrysos

Browse the full AAAI paper archive.