What constitutes a domain for an LLM?
Characterize what constitutes a domain for a large language model using a model-native definition that can be operationalized to study expert specialization and routing behavior, rather than relying on externally imposed human categories.
References
What constitutes a domain for an LLM remains an open question.
— The Expert Strikes Back: Interpreting Mixture-of-Experts Language Models at Expert Level
(2604.02178 - Herbst et al., 2 Apr 2026) in Section 6.2, Experts in the output embedding space
Our domain partition is fixed to four broad domains; learning the partition end-to-end is left to future work.
— DivMoE: Fine-Grained MoE Upcycling via Cross-Domain Expert Composition
(2610.11317 - Lou et al., 8 Oct 2026) in Section A, “Limitations and Future Work”