Kwikinisha-Kuringaniza neza
Kwikinisha wenyine-gutunganya neza bitezimbere icyitegererezo muguhatana cyangwa kwigira kumusaruro wacyo ushize, bikabyara ibimenyetso byamahugurwa.
Incamake
It matters because it can push performance beyond the supervised data using little or no extra human labeling.
Kwibira cyane
Kwikinisha bifite imizi yimbitse mumikino AI: AlphaGo Zero na AlphaZero bageze kumikino ndengakamere bakina imikino miriyoni ubwabo, nta mateka yimikino yabantu. Umwuka umwe ubungubu ugaragara mururimi-rwicyitegererezo neza. Muri SPIN (Kwikinisha-fIne-tuNing), icyitegererezo kigezweho gitanga ibisubizo kubibazo, kandi imyitozo isunika icyitegererezo gutandukanya ibisubizo byabyaye ubwabyo nibya mbere byanditswe n'abantu, bikifata nk'umukinnyi ndetse nuwo bahanganye. Kubisubiramo bikurikiranye 'uwo duhanganye' (igenzura ryabanjirije) birakomera, bityo icyitegererezo kigomba gukomeza gutera imbere, buhoro buhoro kiziba icyuho hamwe nogusaranganya intego. Ubujurire bukomeye nuburyo bukoreshwa neza: imibare ihamye ikurikiranwa irashobora gukanda kugirango ibone inyungu nyinshi idakusanyije imyigaragambyo mishya cyangwa ibyo abantu bakunda.
Ubushishozi
SPIN ikora neza-nkumukino wabakinnyi babiri hamwe nigihombo cya DPO: icyitegererezo cyatojwe gutanga amahirwe menshi kubisubizo byabantu kuruta uko byabyaye ubwabyo uhereye kubisubiramo mbere. Kuberako igenzura ryambere ritanga ibibi, umunzani utoroshye uhita uko moderi itera imbere. Muri sisitemu yo gukina imikino, kwikinisha byahujwe no gushakisha (urugero, MCTS) hamwe numuyoboro wagaciro, bitanga integanyanyigisho zidashira z'abatavuga rumwe na bo buhoro buhoro badafite amakuru yo hanze.
Ingaruka z'Ingamba
Ibyemezo bisobanutse
Iragufasha gutandukanya ibyifuzo bya tekiniki bisobanutse nururimi rwo kwamamaza.
Igiciro na bije
Urashobora kubaza ibibazo byiza byo gushyira mubikorwa mbere yo gukoresha amafaranga cyangwa igihe.
Itsinda hamwe nakazi
Amakipe asangiye ibitekerezo akora ibicuruzwa byiza, politiki, nibyemezo byo kwiga.
Igihe kizaza cyo Kwikinisha Cyiza-Guhuza
Kwikinisha ni umukandida uyobora gusenya urukuta rwamakuru, kubera ko akora integanyanyigisho zayo aho gushingira ku birango byabantu. Tegereza gukura muri domeni zemezwa nkimibare, kode, hamwe na theorem yerekana, aho abagenzuzi bonyine batondekanya kugerageza. Ingaruka zirimo ibihembo bya hacking hamwe nicyitegererezo cyo gusenyuka bivuye mumahugurwa kumusaruro mwinshi cyane, sisitemu rero izaza ishobora guhuza kwikinisha hamwe nibimenyetso bifatika, kugenzura, hamwe nibitekerezo byabantu cyangwa byukuri-isi.
Gushyira mu bikorwa Isi
AlphaGo Zero na AlphaZero bagera kuri superhuman Go, chess, na shogi rwose binyuze mukwikinisha nta mukino wabantu
SPIN izamura amanota ya LLM mugutandukanya byimazeyo ibisubizo byayo nibisubizo byabantu
Imibare na coding byerekana kubyara igisubizo, hanyuma imyitozo kubigenzurwa na cheque yikora cyangwa ibizamini bya unit
Ibiganiro n'ibiganiro bitezimbere ingamba mukinisha inshuro nyinshi ibiganiro byombi ubwabo
Ingaruka & Kurinda
Amakipe atandukanye arashobora gukoresha ijambo rimwe muburyo butandukanye, sobanura intera hakiri kare.
Ibipimo birashobora kugaragara bikomeye mugihe imikorere-yisi-itaringaniye.
Kwirengagiza ubuziranenge bwamakuru na gahunda yo gusuzuma akenshi bitanga ibisubizo byoroshye.
Igishushanyo mbonera
Tangira nururimi rusobanutse rwibisubizo ukeneye.
Toranya intsinzi imwe hamwe nuburyo bumwe bwo gutsindwa mbere yo kwipimisha.
Koresha umuderevu muto hamwe namakuru ahagarariye, ntabwo ari demo yashizweho.
Inyandiko aho Kwikinisha Byiza-Guhuza bifasha kandi nuburyo bworoshye bworoshye.
Komeza Ubushakashatsi
Free newsletter
Get the daily AI briefing
Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.
One email each weekday. Unsubscribe in one click. We never sell or share your address.
Test yourself
Take the Self-Play Fine-Tuning quiz
Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.
Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation
Ubuyobozi bukurikira
Kuringaniza neza
Ibibazo bikunze kubazwa
What is Self-Play Fine-Tuning?
Kwikinisha wenyine-gutunganya neza bitezimbere icyitegererezo muguhatana cyangwa kwigira kumusaruro wacyo ushize, bikabyara ibimenyetso byamahugurwa. Nibyingenzi kuko irashobora gusunika imikorere irenze amakuru yagenzuwe ukoresheje bike cyangwa ntanibindi byanditseho abantu.
Ni ubuhe buryo buzwi bwageze kuri superhuman Go gukina ukoresheje kwikinisha gusa kandi nta nyandiko yimikino yabantu?
AlphaGo Zero yize rwose mumikino yakinnye ubwayo, guhera kumikino idasanzwe no kurenza verisiyo zabanjirije imyitozo yabantu.
Muri SPIN, ni uruhe ruhare rw 'uwo duhanganye' icyitegererezo kigomba gutsinda?
SPIN ifata neza-umukino nkumukino wo kwikinisha aho ibisubizo byabanjirije itera ibyakozwe ubwabyo bikora nkibibi icyitegererezo cyiga kurenga.
Ni izihe nyungu nyamukuru zamakuru-yogukoresha-kwikinisha neza?
Kwikinisha gukora ibimenyetso byayo byamahugurwa, bityo igenzurwa rihamye rishobora gutanga umusaruro ushimishije udakusanyije imyigaragambyo mishya.
Mu ntego ya SPIN yo guhugura, niyihe moderi yasunitswe gukora?
SPIN ikoresha igihombo cyuburyo bwa DPO buhemba gutandukanya ibisubizo byabantu (positif) nibisubizo byambere byashizweho ubwabyo (negatif).
Ni ukubera iki ingorane zo kwikinisha neza-gutunganya ziyongera mu buryo bwikora hejuru yisubiramo?
Kuberako buri cyerekezo kibi kiva muburyo bukomeye bwambere, icyitegererezo gihura ningorabahizi zitoroshye zidafite igishushanyo mbonera.