Diep et al. (2025) establish a theoretical link between the “zero-initialized attention” mechanism and Mixture-of-Experts (MoE), proving that this initialization strategy improves sample efficiency compared to random initialization, with non-linear prompts outperforming linear ones.

Authors

• Person: Not available Organization: arXiv
  A Survey on the Theory and Mechanism of Large Language Models

Sources

• A Survey on the Theory and Mechanism of Large Language Models arxiv.org arXiv via serper

Referenced by nodes (1)

• Mixture of Experts (MoE) concept