Mixture of Experts: Efficient Sparse Models

Fields

id

sources/src5-moe-paper

URL

https://arxiv.org/abs/1701.06538

Publisher

Google

Kind

paper

Published

2017-01-01

Accessed

2026-10-04

notes

Mixture of Experts enables sparse activation of model parameters. Key for scaling LLM capacity without proportional compute cost. Used by DeepSeek, Mixtral, Qwen. Foundation for MoE inference optimization.

Notes

Mixture of Experts enables sparse activation of model parameters. Key for scaling LLM capacity without proportional compute cost. Used by DeepSeek, Mixtral, Qwen. Foundation for MoE inference optimization.

Cross-references

This record references no other records.