Xalaat ci seen xalaat
Li ñuy woowe chaîne de pensée mooy su amee model buy jàngat jafe-jafe bi jéego par jéego ci bind balaa muy joxe tontu bu mujj.
Résumé
This simple change dramatically improves accuracy on math, logic, and multi-step questions.
Plongeur bu xóot
Du nga jubal ci tontu, waaye model chain-of-thought (CoT) dafay bind jéego yu digg yi, lu melni wane sa liggéey ci klaasu math. Benn téere Google bu Jason Wei ak ay naataangoom bind ci atum 2022 dafa wane ni su amee ay model yu mag yu am misaali xalaat yu jéego ci jéego, dafay gëna yokk njariñu liggéey yu metti. Ginaaw loolu yàggul dara, Kojima ak ay naataango gis nañu ni yokk ci 'Nanu xalaat jéego ak jéego' dafay jur xalaat bu amul benn misaal - ñu koy woowe zero-shot CoT. Li gëna am solo mooy njariñ lii mën na feeñ: dafay gëna feeñ ci model yu mag yi te du jàppale yu ndaw yi. Benn setal buñuy woowe méngoo sa bopp dafay jël misaali yooni xalaat yu bari ba noppi jël tontu bi gëna bari, gëna yombal wóor gi.
Gis-gis xarala
Bind jéego yu digg yi dafay may model bi 'espace' bu gëna bari ci xayma - jéego bu nekk buñ defar dafay bokk ci li ñuy dugal ci li ci topp, loolu mooy tax mu xaaj jafe-jafe bu metti ci jéego yu ndaw yu gëna yomba def, moo gën ñuy xalaat ci benn yoon. 2025 vague xeetu xalaat yu melni OpenAI's o-series ak DeepSeek-R1 tabax lii ci saasi: ludul yéem ci ab laaj, ñu tàggat leen ci jàng buy dooleel ngir génne chaine yu guddu ci biir xalaat, banneexu, saytu,. R1 dafa wane bu baax ni xalaat mën na génn ci RL bu sell.
njeextalu pexe
Gaawaay ak yaatuwaay
Liggéeyukaay yi ci làkk yi mën nañu gëna gaaw te duñu yàq deggoo gi.
Dugg ak yegg
Dafay yaatal jëfandikoo gi ci làkk yi ak ci anam yi ñuy jokkoo.
dogal yu gëna leer
Ekip yi mën nañu gëna yàgg ci àtte ci jamono ji otomatisation di liggéey ci baamtu.
Ëlëgu xalaat bu lalu ci xalaat
Caden de pensée joge na ci ab pexe buy yee nit ñi mujjee nekk paradigm buy tàggat nit. Xaarandi yeneen 'model xalaat' yuy dugal xaalis bu gëna bari ci inference - li ñuy woowe test-time compute - gaawaayu njaay ngir gëna jubal ci jafe-jafe yu metti, ak niveau effort yuñ mëna regle. Laaj yi ubbeeku ñooy ndax chaine bind bi dafay wane bu baax anam wi model bi di doxee, naka lañu mëna moytu xalaat yu yàgg yi jur njuumte, ak naka lañu mëna yemale njëg yi. Kalite xalaat, du xam-xam bu bees rek, mooy nekk axis bi gëna mag bi model yu mag yi di joŋante.
Doxal ci àdduna dëgg
Faj jafe-jafe baat math yu bari jéego ci teg jéego arithmetik bu nekk balaa nimero bu mujj bi.
Kodu debugging ci xalaat li ligne bu nekk di def ak fi logique bi di dagg.
Tontu ay puzzle logique wala waajal ay liggéey yuy laaj topp ay constraints yu bari benn yoon.
Jëfandikoo deggoo sa bopp ngir jël misaali yooni saafara yu bari te tànn tontu li gëna bari ci laaj bu jafee xam.
Risk yi ak balustrade yi
Lépp lu jaarul yoon mën na dugg ci rapoor yi, jàppale ci liggéey bi, wala ci njariñu gëstu bi.
Sensibilite bu gaaw mën na jur njariñ yu wuute ci laajte yu noonu mel.
Done yu am solo mën nañu feeñ sudee seytu jëfandikoo gi néew doole.
Roadmap ngir samp gi
Mandargal formaa génne gi, melokaan bi, ak standard kalite yi laata ngay dugal ko.
Tontu yu am solo ak balluwaay yu wóor saa yu dëggu bi di am solo.
Fexeel am barabu xool nit ñi ngir am njariñ yu am solo.
Toppal anami gacce yi ak di faral di tàggataat ay laaj wala def-liggéey.
Weyal di banneexu
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 Chain-of-Thought Reasoning 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
Gis bi ci topp
Garab-ci-Xalaat Xalaat
Laaj yi ñuy faral di laaj
What is Chain-of-Thought Reasoning?
Li ñuy woowe chaîne de pensée mooy su amee model buy jàngat jafe-jafe bi jéego par jéego ci bind balaa muy joxe tontu bu mujj. Coppite bu yomb bii dafay gëna baaxal njubte ci math, logique ak laaj yu bari jéego.
Luy xalaat bu lalu ci xalaat?
Chain-of-thought dafay tax model bi wane liggéeyam jéego par jéego, loolu mooy gëna rafetal njub ci jafe-jafe yu bari jéego.
Ban frase bu yomb lañu wane ni dafay indi xalaat jéego ci jéego te amul misaal (zero-shot CoT)?
Kojima ak ñeneen. (2022) dafa wane ni yokk 'Nanu xalaat jéego par jéego' dafay tax ñu xalaat doonte amul misaal yuñ liggéey.
Luy pexem méngoo sa bopp?
Teg sa bopp dafay jur xalaat yu bari yu moom seen bopp ba noppi jël wote yu ëpp ci tontu bi, gëna yombal wóor gi.
Lan moo wuutale xeetu xalaat yu jamonoy 2025 yu melni o-serie AIU_PROTECTED_10 ak DeepSeek-R1 ak laaj CoT bu leer?
Royuwaayi xalaat yooyu dañuy dugal xalaat jéego-ci-jéego ci model bi jaaraleko ci tàggat (espesialeman RL), suko defee ñu xalaat te soxla wuñu benn laaj bu amul benn werante saa yu nekk.
Lan moo waral bind jéego yu digg yi di gëna yombal tontub model bi?
Jéego bu nekk buñu bind dafay nekk contexte ci li ci topp, may model bi palaas ngir dindi jafe-jafe bi ci barabu xalaat ci benn yoon - waaye loolu garantiwul ni dina jub.