ScholarGate
Assistent
Machine learning

Teadmise destilleerimine

Teadmise destilleerimine on mudelite tihendamise tehnika, mille võtsid 2015. aastal kasutusele Geoffrey Hinton ja kolleegid. See treenib väikest õpilas-mudelit suure õpetaja-mudeli pehmete väljundite abil. Destilleeritud mudelid nagu DistilBERT ja TinyBERT saavutavad umbes 97% suurema mudeli jõudlusest, töötades samal ajal palju kiiremini.

Ava rakenduses MethodMindPeagiVideoPeagiDownload slides

Loe meetodi täielikku kirjeldust

Ainult liikmetele

Selle osa lugemiseks logi sisse tasuta kontoga.

Logi sisse

Method map

The neighbourhood of related methods — select a node to explore.

+3 more

Allikad

  1. Hinton, G., Vinyals, O. & Dean, J. (2015). Distilling the Knowledge in a Neural Network. NeurIPS Deep Learning Workshop. link
  2. Sanh, V., Debut, L., Chaumond, J. & Wolf, T. (2019). DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter. arXiv:1910.01108. link

Kuidas sellele lehele viidata

ScholarGate. (2026, June 1). Knowledge Distillation (Teacher–Student Model Compression). ScholarGate. https://scholargate.app/et/deep-learning/knowledge-distillation

Which method?

Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.

Compare side by side

Sellele viitavad

ScholarGateKnowledge Distillation (Knowledge Distillation (Teacher–Student Model Compression)). Loetud 2026-06-15 aadressilt https://scholargate.app/et/deep-learning/knowledge-distillation · Andmestik: https://doi.org/10.5281/zenodo.20539026