Ururimi AI
Ubushobozi bwihutirwa bwururimi runini
Ubushobozi bwihuse nubuhanga bugaragara gitunguranye mururimi runini iyo rumaze gutsinda igipimo runaka, nubwo moderi ntoya iterekanye ikimenyetso cyayo.
Umva iki gitaboAI-byavuzwe · ~2 min
Kuriyi page2 min soma
Incamake
Bifite akamaro kuko bakora ubushobozi bigoye guhanura uhereye kubushakashatsi buto.
Kwibira cyane
Yamamaye mu mpapuro 2022 na Wei na bagenzi be, kugaragara bivuga imirimo aho imikorere iguma hafi yamahirwe kuri moderi ntoya hanyuma igasimbuka cyane iyo icyitegererezo cyambutse ingano yubunini mubipimo, amakuru, cyangwa kubara. Ingero zavuzwe zirimo intambwe nyinshi zo kubara, ibipimo bimwe byo gutekereza, hamwe n'amabwiriza mashya. Igice cyagaragaye ni uguhagarika umutima: ubuhanga ntabwo bwagendaga butera imbere buhoro buhoro, byasaga nkaho bidahari hanyuma bihari. Ikurikiranwa rya 2023 ryakozwe na Schaeffer na bagenzi be bavuze ko kugaragara ko ari bimwe mu bikoresho byo gupima, kubera ko ibipimo bikaze byose cyangwa nta na kimwe nko guhuza neza gukabya gusimbuka gutunguranye bisa neza neza no gutsinda amanota yoroshye. Impaka zahinduye uburyo abashakashatsi batanga ibisubizo bipima no gutora ibipimo byo gusuzuma.
Gushyira mu bikorwa Isi
Ingero nini zikemura ibibazo byinshi byintambwe ijambo ibisubizo bito byashubijwe kurwego rwamahirwe.
Icyitegererezo gitunguranye gikurikira ibintu bigoye, ntabwo-byigeze biboneka amabwiriza nyuma yo kurenga igipimo.
Urunigi-rwibitekerezo rutera imbaraga zo gutekereza gusa iyo moderi igeze mubunini buhagije.
Abashakashatsi bongeye gutegura igipimo gitunguranye 'gitunguranye' hamwe no gutanga amanota igice-cy'inguzanyo no kubona umurongo mwiza.
Kazoza k'ubushobozi bwihuse bwururimi runini
Abashakashatsi ubu bahuza ubushakashatsi bwo gupima hamwe na metero nyinshi kugirango batandukane impinduka zukuri nukuri, kandi bareba ubushobozi bugera mubipimo gusa. Ibyiza byo guhanura bifite umutekano, kubera ko ubushobozi butunguranye bushobora kubamo ibyago. Tegereza imirimo myinshi yo gupima amategeko ateganya ubushobozi hakiri kare, hiyongereyeho igishushanyo mbonera cyitondewe kuburyo byavuzwe ko 'kugaragara' byerekana imyitwarire yicyitegererezo aho kuba igipimo cyo gupima.
Kubishyira mubikorwa
Impamvu ari ngombwa
- Umuvuduko n'igipimo. Ururimi rwakazi rushobora kugenda byihuse nta gutamba guhuzagurika.
- Kugera no kugera. Yagura uburyo bwindimi nuburyo bwo gutumanaho.
- Ibyemezo bisobanutse. Amakipe arashobora kumara umwanya munini murubanza mugihe automatike ikora gusubiramo.
Witondere
- Ibintu bifatika bishobora kwinjiza bucece raporo, gushyigikira imigendekere, cyangwa ibisubizo byubushakashatsi.
- Kwihuta byihuse birashobora gukora ibisubizo bidahuye mubisabwa bisa.
- Ibyanditswe byumvikana birashobora kugaragara niba kugenzura kugenzura ari ntege.
Intambwe ikurikira
- Sobanura imiterere isohoka, amajwi, hamwe nubuziranenge mbere yo gutangira.
- Ibisubizo byibanze hamwe nisoko yizewe igihe cyose ukuri kwingirakamaro.
- Komeza kugenzura abantu kugenzura ibisubizo byinshi.
- Kurikirana uburyo bwo kunanirwa no kongera imyitozo cyangwa akazi gahoraho.
Komeza Ubushakashatsi
Ibaruwa y'amakuru y'ubuntu
Ibuka amakuru ya buri munsi ya AI
Inkuru eshatu zizewe za AI buri gitondo cy'iminsi y'akazi, zanditse mu rurimi rworoshye. Ubuntu iteka ryose, nta kwamamaza.
Imeri imwe buri munsi y'akazi. Kwiyimura mu buryo bumwe. Ntitwigera tugurisha cyangwa ngo dusangire aderesi yawe.
Isuzume
Take the Emergent Abilities of Large Language Models quiz
Ibisubizo ako kanya kuri buri kibazo, n'icyemezo gishobora gusangirwa gifite ID yagenzurwa umaze gutsinda isomo.
Fasha uburezi bwa AI bw'ubuntu. AI Understanding ni umuryango utegamiye kuri leta wa 501(c)(3) — nta kwamamaza, nta kwishyura, iteka ryose. Tanga impano
Ibibazo bikunze kubazwa
Ni ubuhe bushobozi bwihutirwa bwururimi runini?
Ubushobozi bwihuse nubuhanga bugaragara gitunguranye mururimi runini iyo rumaze gutsinda igipimo runaka, nubwo moderi ntoya iterekanye ikimenyetso cyayo. Bifite akamaro kuko bakora ubushobozi bigoye guhanura uhereye kubushakashatsi buto.
What did early emergence reports describe?
The cited work described apparently sharp capability changes for some tasks.
What alternative did Schaeffer and colleagues argue?
The follow-up argues nonlinear or discontinuous metrics can create apparent abrupt jumps.
Why can exact-match scoring make a gradual skill look abrupt?
The study explains that all-or-nothing metrics can exaggerate a discontinuity.
What might happen if partial-credit or continuous metrics are used?
The paper reports metric changes can reveal smoother performance trends.
What is needed before concluding a new ability appeared at a scale threshold?
The research debate emphasizes metric choice and better statistics.
Komeza wige
Kuyobora
Abandi bayobozi batoranijwe kuriyi ngingo