Analogique ak jéego-dellu ginaaw
Analogical ak Step-Back prompting ñoom ñaar ñooy teg model bi mu njëkka xalaat ci niveau bu gëna kawe: analogical prompting daf koy fàttaliku jafe-jafe yuñ saafara, waaye step-back daf koy génne ci njàngale mi laata muy wax ci ay mbir yu gëna yomb.
Résumé
They matter because abstraction often beats diving straight into details.
Plongeur bu xóot
Lii ñaari pexe gëstu yu jëm ci 2023 ngir gëna mëna xalaat. Analogical prompting, bi Yasunaga ak ay naataangoom def, dafay laaj model bi mu defar boppam yenn misaal yu am solo, jafe-jafe yu noonu mel yi mu gis bu baax, ak seeni pexe balaa ñuy saafara jafe-jafe biñ bëgga, dindi soxla misaal yuñ bind ak loxo. Jéego-Dellu ginaaw, bi Zheng ak ay naataangoom def, lu moy loolu dafay njëkka laaj ab abstraction ("ban njàngale bu ëpp solo wala dëgg moo jiite lii?"), jëlaat wala sabab ci njàngale boobu, ba noppi jëfandikoo ko ci laaj bu konkret bi. Ñoom ñaar ñépp dañuy puus model bi ci detay yu teel. Step-Back dafa wane njariñ ci laaji physique ak chimie ak ci xalaat yu bari, ci noonu la analogique di gëna yokk math ak defar kode ci méngale misaal yi ak jafe-jafe bu nekk.
Gis-gis xarala
Jéego-dellu ginaaw dafay dox ndax grounding tontu ci benn yoon buñu wax (maanaam, yoon gaz ideal wala benn firnde) dafay tënk xalaat yu leer yi ci topp ba noppi wàññi slips ci diggante jéego yi. Analogique prompting dafay dox ndax misaal yi defar seen bopp dañu méngoo ak jafe-jafe bi ñuy wax, lu bari luñu gëna am solo misaal yu néew yu fiks, te dañuy njëkka am motif solution bu méngoo. Ñoom ñaar ñu ngi soppi calcul bi ngir njëkka am abstraction bu dëggu bi, ginaaw ga ñu def liggéey bu leer te leer.
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 Analogique ak Jéego-Dellu ginaaw
Expect these abstraction-first patterns to merge with retrieval systems, where the step-back principle becomes a precise query into a knowledge base, and with agent planners that reason about strategy before tactics. Gëstu dafay setal sudee analogie yi defar seen bopp dañuy jàppale ak sudee dañu dugal misaal yu amul njariñ wala yu jaarul yoon, ba noppi boole jéego yu dellu ginaaw ak xool ndax ñu xool njàngale miñ tànn balaa ñuy tabax ci kaw xalaat yu diis. Daa melni dañuy nekk default ci model yuñ defaree xalaat.
Doxal ci àdduna dëgg
Sooy tontu benn laaj ci wàllu physique, dangay njëkka wax yoon wi ci laale (lu melni, ñaareelu yoonu Newton) jaare ko ci jéego yu ginaaw, ba noppi nga boole ci ay lim
Faj ab jafe-jafe math bu bees ci fexe ba model bi fàttaliku ñaari jafe-jafe yuñ saafara nuru ci laaj analogique
Xoolaat ab laaj bu bari ay hop ci dellu ci kategori wala entite bu gëna yaatu balaa ngay boole ay mbir
Defar kode ci defar sa bopp algorithm analogue ak pexem, ba noppi méngale ko ak liggéey bi fi nekk
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 Analogical and Step-Back Prompting 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
Li gëna néew-ba-li gëna am solo
Laaj yi ñuy faral di laaj
What is Analogical and Step-Back Prompting?
Analogical ak Step-Back prompting ñoom ñaar ñooy teg model bi mu njëkka xalaat ci niveau bu gëna kawe: analogical prompting daf koy fàttaliku jafe-jafe yuñ saafara, waaye step-back daf koy génne ci njàngale mi laata muy wax ci ay mbir yu gëna yomb. They matter because abstraction often beats diving straight into details.
Lan la analogique prompting di sàkku ci model bi balaa muy saafara jafe-jafe biñ bëgga?
Analogique prompting moo waral model bi di defar ay misaal ak ay pexe yu méngoo ak jafe-jafe bi.
Lan mooy jéego bu njëkk bi nga wara def sooy laaj ñu dellu ginaaw?
Step-Back dafay njëkka laaj ab laaj abstraction ngir wane njàngale mi wala liy doxal jafe-jafe bi.
Lan moo waral tontub tontu ci njàngale buñ wax (Step-Back) di jàppale?
Wax njàngale mi jiite mooy njëkka tegtal ak tënk jéego yu leer yi, wàññi njuumte yi.
Ci ban xeetu liggéey la Step-Back wane njariñ yu am solo?
Jéego-dellu ginaaw gëna mëna def ci laaji science ak xalaat yu bari ci abstraction bu njëkk.
Lan mooy risk biñ gëna xam ci laaj analogique?
Ndax model bi dafay sosal ay analogie boppam, yenn saa misaal yooyu mën nañu baña am njariñ wala ñu juum, te loolu mën na gaañ tontu li.