ScholarGate
Assistente

Comparar métodos

Examine os métodos selecionados lado a lado; as linhas que diferem ficam destacadas.

Destilação de Conhecimento×Mistura de Especialistas×
ÁreaAprendizado profundoAprendizado profundo
FamíliaMachine learningMachine learning
Ano de origem20152017
Autor originalHinton, G., Vinyals, O. & Dean, J.Shazeer, N. et al.
TipoNeural network compression (teacher–student)Sparse neural network architecture (conditional computation)
Fonte seminalHinton, G., Vinyals, O. & Dean, J. (2015). Distilling the Knowledge in a Neural Network. NeurIPS Deep Learning Workshop. link ↗Shazeer, N. et al. (2017). Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer. ICLR. arXiv:1701.06538 link ↗
Outros nomesBilgi Damıtma (Knowledge Distillation), bilgi damıtma, teacher-student distillation, model distillationUzman Karışımı (Mixture of Experts — MoE), uzman karışımı, MoE, sparse mixture of experts
Relacionados53
ResumoKnowledge Distillation is a model-compression technique, introduced by Geoffrey Hinton and colleagues in 2015, that trains a small student model using the soft-label outputs of a large teacher model. Distilled models such as DistilBERT and TinyBERT reach roughly 97% of the larger model's performance while running far faster.Mixture of Experts (MoE) is a sparse neural-network architecture, introduced by Shazeer and colleagues in 2017 with the sparsely-gated MoE layer, in which only a subset of expert sub-networks is activated for each input. As seen in models such as Switch Transformer and Mixtral, it holds computation cost fixed even as the total parameter count grows.
ScholarGateConjunto de dados
  1. v1
  2. 2 Fontes
  3. PUBLISHED
  1. v1
  2. 2 Fontes
  3. PUBLISHED

Ir para a pesquisa Baixar slides

ScholarGateComparar métodos: Knowledge Distillation · Mixture of Experts. Recuperado em 2026-06-18 de https://scholargate.app/pt/compare