Ururimi AI

Kugenzura Inzira Kubitekerezo

Igenzura ryimikorere rihemba icyitegererezo kuri buri ntambwe ikwiye murwego rwo gutekereza, ntabwo igisubizo cyanyuma.

2 min somaIbiherutse kuvugururwa

Incamake

For math, where one wrong move ruins everything, grading the work itself produces far more reliable solvers.

Kwibira cyane

Ingero nyinshi zerekana ibihembo zitanga igisubizo cyanyuma (kugenzura ibisubizo). Ibyo bituma icyitegererezo 'gira amahirwe' - kugera ku mubare ukwiye binyuze mu ntambwe zitagira inenge zihagarika. Igenzura ryibikorwa ahubwo ritoza uburyo bwo guhemba inzira (PRM) kubirango byabantu cyangwa AI biranga buri ntambwe yo hagati nkukuri, itariyo, cyangwa itabogamye. __AIU_PROTECTED_10 __ '2023' Reka dusuzume intambwe ku yindi 'impapuro zasohoye PRM800K, hafi 800.000 byurwego rwintambwe kurwego rwibibazo bya MATH, kandi rwerekana igenzurwa ryakozwe nigenzura ryakemuye 78% byikizamini cyibizamini hamwe n’ibisubizo bidakomeye gusa. PRM ikoreshwa muburyo bwo gutondekanya ibisubizo byinshi byatoranijwe, gutoranya urunigi hamwe namanota ntarengwa yo gutera intambwe. Iratanga kandi ibitekerezo byumvikana: urashobora kubona neza aho ibitekerezo bitandukiriye.

Ubushishozi

Mugihe cyibizamini icyitegererezo cyerekana ibisubizo byinshi byabakandida; PRM itanga amanota kuri buri ntambwe kandi igisubizo cyibisubizo muri rusange nibicuruzwa (cyangwa byibuze) byintambwe imwe yo gukosorwa. 'Ibyiza-bya-N' noneho hitamo urunigi rwo hejuru. Kuberako inguzanyo yatanzwe mugace, ibimenyetso byamahugurwa birushijeho kuba urusaku kandi bitaranguruye urusaku kurenza igihembo cyanyuma-cyikurikiranya, bigabanya ibihembo-hacking aho intambwe mbi zahuriranye zitanga ibisubizo nyabyo.

Ingaruka z'Ingamba

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.

Ejo hazaza h'ubugenzuzi bwibikorwa byo Gutekereza Imibare

Intambwe yintoki irazimvye, ubushakashatsi rero buhindukira kugenzura kugenzura byikora - ukoresheje Monte Carlo umuzingo (Math-Shepherd) kugirango ugereranye agaciro ka buri ntambwe nta kirango cyabantu, cyangwa kugira moderi zikomeye zicira intege nke. Tegereza PRMs gutwara imbaraga-yiga-gushimangira neza, ntabwo ari ugusubiramo gusa, no gukwirakwiza imibare mu kode, ibimenyetso bya siyansi, hamwe na gahunda yo gutangiza intambwe nyinshi aho gukosora intambwe kurwego.

Gushyira mu bikorwa Isi

__AIU_PROTECTED_10 __ 'PRM800K dataset: 800K ibirango byintambwe yumuntu byifashishwa mugutoza abagenzuzi kubipimo bya MATH

Imibare-Umwungeri: ihita yerekana intambwe ikosowe ikoresheje Monte Carlo kugirango wirinde gutangaza abantu bihenze

Ibyiza-bya-N reranking: gutanga ibisubizo 256 no guhitamo imwe PRM itanga amanota menshi kuri buri ntambwe

Ibikoresho byo kwigisha byerekana umurongo nyawo mubisubizo byakozwe numunyeshuri aho ikosa rigaragara bwa mbere

Ingaruka & Kurinda

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.

Igishushanyo mbonera

1

Sobanura imiterere isohoka, amajwi, hamwe nubuziranenge mbere yo gutangira.

2

Ibisubizo byibanze hamwe nisoko yizewe igihe cyose ukuri kwingirakamaro.

3

Komeza kugenzura abantu kugenzura ibisubizo byinshi.

4

Kurikirana uburyo bwo kunanirwa no kongera imyitozo cyangwa akazi gahoraho.

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 Process Supervision for Math Reasoning quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

Tangira ikibazo

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

Ubuyobozi bukurikira

Urunigi-rw'ibitekerezo

Ibibazo bikunze kubazwa

What is Process Supervision for Math Reasoning?

Igenzura ryimikorere rihemba icyitegererezo kuri buri ntambwe ikwiye murwego rwo gutekereza, ntabwo igisubizo cyanyuma. Kubiharuro, aho kwimuka nabi kwangiza ibintu byose, gutondekanya akazi ubwabyo bitanga ibisubizo byizewe cyane.

Ni irihe tandukaniro nyamukuru riri hagati yo kugenzura inzira no kugenzura ibizagerwaho?

Igenzura ryibikorwa ritanga ikimenyetso cyigihembo kuri buri ntambwe yo gutekereza, mugihe ubugenzuzi bwibisubizo bugenzura igisubizo cyanyuma.

Ni ukubera iki kugenzura ibisubizo gusa bishobora kuyobya ibibazo by'imibare?

Guhemba igisubizo cyanyuma gusa inguzanyo 'amahirwe' ibisubizo aho intambwe yo hagati itari yo ihurirana itanga ibisubizo nyabyo.

Niki OpenAI yasohoye hamwe nakazi ka 'Reka dusuzume intambwe ku yindi'?

PRM800K ikubiyemo ibisobanuro hafi 800.000 byabantu byerekana intambwe yo gutekereza kugiti cye nkukuri cyangwa atari byo.

Nigute uburyo bwo guhemba uburyo bukoreshwa mugihe cyo gufata umwanzuro?

PRM itanga amanota kuri buri ntambwe yibisubizo byinshi byabakandida, igafasha ibyiza-N-guhitamo amanota menshi yo gutekereza.

Nubuhe buryo bugabanya gukenera ibirango bihenze byabantu?

Imibare-Umwungeri igereranya intambwe ikosorwa mugutangiza ibyuzuye no gupima inshuro bagera kubisubizo nyabyo, birinda kuranga intoki.