Xiga xigaHagaha xiga
Distillation Diffusion Adversarial iyo Turbo Models
Farsamo
HAGAHA Farsamada
Faafinta Aqoonta
Knowledge distillation trains a student model using a teacher model’s outputs or representations as supervision.
Dhageyso hagahanAI-waa laga sheekeeyay · ~2 min
Boggaan2 daqiiqo akhri
Dulmar
It can support more efficient deployment, but the student’s accuracy and behavior depend on its capacity, training data, and the teacher signal.
quusid qoto dheer
Big models are accurate but slow and expensive to deploy. Knowledge distillation transfers their capability into a compact model by having the student learn from the teacher's outputs rather than only from hard labels. The key insight, from Hinton and colleagues, is that a teacher's full probability distribution carries 'dark knowledge': even when it predicts 'dog', the relative probabilities for 'wolf' versus 'car' reveal how the teacher sees similarities. Softening these probabilities with a temperature exposes that structure, and the student is trained to match it, often alongside the true labels. The goal is an efficient student that benefits from teacher supervision; the actual accuracy and speed must be measured for the task and implementation. DistilBERT and TinyBERT are well-known distilled language models.
Dhaqangelinta Adduunka-dhabta ah
DistilBERT waxay ku cadaadinaysaa BERT qiyaas ahaan 40% ka yar iyada oo la ilaalinayo fahamka luqadda inteeda badan si loo ogaado deg degga.
Yaraynta qaabka aragga weyn si sawir-kala-soocida sawirku ugu shaqeeyo wakhtiga dhabta ah abka kamarada casriga ah.
U kala qaybinta fikradda silsilad-fikirka ee model weyn qaab ka yar si ay uga jawaabto xisaabta ama su'aalaha kood ka jaban.
Isku-dubbaridka moodooyinka hal arday si wax-soo-saarka u adeego kharashyada iyo hoos u dhaca daahitaanka iyada oo aan la lumin saxnaansho badan.
Mustaqbalka Faafinta Aqoonta
Distillation hadda waa tillaabada caadiga ah ee dhoofinta moodooyinka hufan waxayna udub dhexaad u tahay mowjadaha maanta ee moodooyinka yaryar ee karti leh. Isbeddellada degdega ah u koraya waa isku xigxig-is-daba-marin laga soo bilaabo moodooyinka luqadaha waaweyn, halkaas oo qaabka xooggani uu ka soo saaro xogta tababarka ama raad-raacyada sababaynta (oo ay ku jiraan silsilad-fikirka) si loo baro ardayda yaryar, oo ku kharribisa khadka xogta synthetic. Filo isku-duubnaan adag oo leh qiyaasid iyo manjooyin, qalab geyn badan, iyo dood joogto ah oo ku saabsan shatiga iyo tayada marka aad ka fogaato moodooyinka lahaanshaha kuwaas oo wax soo saarkoodu ay noqdaan calaamada tababarka tartame.
Ku dhaqangelinta
Maxay muhiim u tahay
- Qiimaha iyo miisaaniyada. Go'aamada qaab-dhismeedku waxay horseedaan waxqabadka iyo kharashka hawlgalka sannadaha.
- Go'aamo cad. Waxbarashada farsamada waxay ka caawisaa kooxaha inay doortaan xidhmo sax ah, ma aha oo kaliya kan ugu cusub.
- Xakamaynta tayada. Doorashooyinka injineernimada ee wanaagsan waxay yareeyaan shilalka la isku halleyn karo ee wax soo saarka.
Iska ilaali
- Hagaajinta hal bartilmaameed waxay qarin kartaa daciifnimada nidaamka ballaaran.
- Kaabayaasha dhaqaalaha iyo dayactirka inta badan waa la dhayalsadaa.
- Nabadgelyada iyo daldaloolada u fiirsashada ayaa kori kara marka nidaamyadu noqdaan kuwo aad u adag.
Tallaabooyinka xiga
- Qeex daahida, tayada, iyo bartilmaameedyada qiimaha ka hor inta aan la hirgelin.
- Benchmark marka la eego culeyska dhabta ah iyo xaaladaha xogta.
- La socodka qalabka khaladaadka, leexashada, iyo saamaynta isticmaalaha.
- U diyaari dib-u-noqoshada iyo dariiqyada jawaab-celinta dhacdada ka hor inta aanad miisaan.
Sii wad Sahaminta
Warqad bilaash ah
Hel warbixinta maalinlaha ah ee AI
Saddex sheeko oo la xaqiijiyay oo AI ah maalin kasta oo shaqo ah, oo lagu qoray Ingiriis fudud. Bilaash had iyo jeer, aan xayaysiis lahayn.
Hal iimayl maalin kasta oo shaqo ah. Ka bax hal gujin. Marna ma iibinno ama ma wadaagno ciwaankaaga.
Is tijaabi
Take the Knowledge Distillation quiz
Fikrad degdeg ah oo ku saabsan jawaab kasta, iyo shahaado la wadaajin karo oo leh aqoonsi la xaqiijin karo markaad ka gudubto koorso.
Taageer waxbarashada AI bilaashka ah. AI Understanding waa 501(c)(3) aan faa'iido doon ahayn — xayaysiis la'aan, albaab lacag ah ma jiro, had iyo jeer. Samee deeq
Su'aalaha soo noqnoqda
What is Knowledge Distillation?
Knowledge distillation trains a student model using a teacher model’s outputs or representations as supervision. It can support more efficient deployment, but the student’s accuracy and behavior depend on its capacity, training data, and the teacher signal.
In knowledge distillation, what is the student model trained to do?
Distillation waxay u beddeshaa kartida macalinka weyn qaab arday ah oo isku dhafan, dhaqso badan.
Maxaa loola jeedaa 'aqoonta madow' ee macalinka?
Aqoonta mugdiga ahi waa qaab-dhismeedka qaybinta ixtimaalka buuxda ee macalinka, sida sida ay 'yeey' ula mid tahay 'ey', ma aha oo kaliya jawaabta sare.
Doorkee ayuu heerkulku (T) ka ciyaaraa distillation?
Heerkulka sare waxa uu balaadhiyaa qaybinta ixtimaalka, taas oo muujinaysa isku ekaanshaha ardaygu ay tahay in uu barto.
Khasaare ka soo baxa caadiga ah wuxuu isku daraa labada qaybood?
Ardaygu waxa uu la mid yahay qaybinta jilicsan ee macalinka (KL) halka uu sidoo kale ku haboon yahay summadaha runta-hoose (cross-entropy).
Keebaa tusaale u ah qaabka luqadda distilled?
DistilBERT waa moodel caan ah oo laga soocay BERT, isagoo haya inta badan waxqabadka oo leh xaddi yar.
Sii wad waxbarashada
Tilmaamaha la xidhiidha
Tilmaamayaal badan ayaa loo doortay mawduucan