Jing Liu

It is found that expert selection is far more redundant than the field's operating points assume, and how much expert computation fine-grained MoEs can dispense with is revealed, and when dynamic allocation earns its complexity is established, informing both practical deployment and future pruning methods.

Structural alignment—not affective warmth or semantic relevance—is the primary conversational feedback mechanism driving grammatical acquisition in sample-constrained language models.

Retrieval-augmented language models can significantly narrow the lexical frequency gap in syntactic contrast sensitivity, but they still leave room for improvement.