Jagoran Harshe AI

Samfuran Harshe

Samar da harshe shine aiki mai sauƙi na yaudara na tsinkayar abin da kalma ko alama ta zo gaba, idan aka ba da rubutu ya zuwa yanzu.

2 min karatuAn sabunta ta ƙarshe

Dubawa

This single objective, scaled up massively, is what produces today's powerful chatbots and writing assistants.

Zurfafa nutsewa

A ainihinsa, ƙirar harshe yana ba da yuwuwar zuwa jeri na rubutu. Ganin cewa 'Babban birnin Faransa shine', yana ƙididdige yadda kowane alamar alama ta gaba zata kasance, kuma 'Paris' ya kamata ya ci sosai. Samfuran harshe na farko sune n-gram na ƙididdiga waɗanda kawai ke ƙididdige sau nawa jerin kalmomi suka bayyana, amma sun yi gwagwarmaya da dogon mahallin da kalmomin da ba a gani. Samfuran yaren jijiya sun maye gurbin kirgawa tare da wasiƙun da aka koya, da kuma gine-ginen gidan wuta daga 2017 ya bar samfura su halarci dogon zangon rubutu da inganci. Manyan yare na zamani kamar dangin GPT an horar da su akan babban rubutun corpora tare da manufa ɗaya: tsinkaya alama ta gaba. Abin sha'awa shine, yin wannan da kyau yana tilasta abin ƙira don ɗaukar nahawu, gaskiya, tsarin tunani, da salo, domin tsinkayar rubutu daidai yana buƙatar fahimtarsa. Generation yana aiki ta maimaita tsinkayar alamar gaba da sake ciyar da shi.

Fahimtar Fasaha

Yawancin nau'ikan yare na zamani suna jujjuya kai: suna ƙididdige yuwuwar jumla zuwa samfur na yuwuwar alama ta gaba, tsinkaya alama ɗaya a lokaci hagu zuwa dama. Horon yana rage asarar giciye-entropy, wanda ke ba da lada mai ƙima ga ainihin alamar gaba a cikin rubutun horo. Wannan abin kulawa ne da kansa, alamun suna zuwa kyauta daga rubutun kansa, don haka ba a buƙatar bayanin ɗan adam. A lokacin tsara, dabarun samfur kamar zafin jiki, top-k, da top-p (nucleus) suna sarrafa ciniki tsakanin abin da ake iya faɗi da fitarwa.

Dabarun Tasiri

Gudu da sikelin

Gudun aikin harshe na iya tafiya da sauri ba tare da sadaukar da daidaito ba.

Shiga ku isa

Yana faɗaɗa damar shiga cikin harsuna da salon sadarwa.

Shawarwari masu haske

Ƙungiyoyi za su iya ciyar da ƙarin lokaci akan hukunci yayin da aiki da kai ke sarrafa maimaitawa.

Makomar Samfuran Harshe

Hasashen alama na gaba ya tabbatar da ƙarfi mai ban mamaki, kuma dokokin ƙididdiga sun nuna cewa manyan ƙira da ƙarin bayanai suna ci gaba da haɓaka iyawa, duk da cewa ribar tana raguwa kuma bayanai masu inganci suna ƙaranci. Iyakar tana jujjuyawa zuwa tunani, doguwar tagogi na mahallin, da hanyoyin horarwa kamar ƙarfafa koyo daga ra'ayin ɗan adam wanda ke siffanta ɗabi'a bayan an gina ƙirar tushe. Yi tsammanin ci gaba da haɗuwa da ƙirar harshe tare da kayan aiki, dawo da bayanai, da abubuwan shigar da abubuwa da yawa, yayin da ainihin maƙasudin tsinkaya-na gaba-alamu ya kasance tushen duk abin da aka gina a kai.

Aiwatar da Gaskiyar Duniya

Cika atomatik a madannai na wayarku ko imel wanda ke ba da shawarar kalma ta gaba yayin da kuke bugawa

Botbot kamar ChatGPT yana samar da cikakkiyar amsa ta maimaita tsinkayar alama ta gaba.

Masu gyara lamba kamar GitHub Copilot suna tsinkayar layin lamba na gaba daga mahallin da ke kewaye

Tsarukan tantance magana ta amfani da samfurin harshe don zaɓar mafi kyawun kwafi a tsakanin zaɓuɓɓukan sauti iri ɗaya

Hatsari & Tsare-tsare

Abubuwan da aka ruɗe suna iya shigar da rahotanni cikin nutsuwa, kwararar tallafi, ko abubuwan bincike.

Hankali na gaggawa na iya ƙirƙirar sakamako mara daidaituwa a cikin buƙatun iri ɗaya.

Za a iya fallasa bayanan rubutu mai ma'ana idan ikon samun dama yana da rauni.

Taswirar Hanya

1

Ƙayyade tsarin fitarwa, sautin, da ma'auni masu inganci kafin fitowa.

2

Amsa a ƙasa tare da amintattun tushe a duk lokacin da daidaito ya shafi mahimmanci.

3

Ajiye wurin binciken ɗan adam don abubuwan da ake samu masu girma.

4

Bibiyar tsarin gazawar kuma sake horar da tsokaci ko tafiyar aiki akai-akai.

Ci gaba da Bincike

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 Language Modeling quiz

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

Fara tambayoyi

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

Jagora na gaba

Model Harshen Marufi

Tambayoyin da ake yawan yi

What is Language Modeling?

Samar da harshe shine aiki mai sauƙi na yaudara na tsinkayar abin da kalma ko alama ta zo gaba, idan aka ba da rubutu ya zuwa yanzu. Wannan makasudi guda ɗaya, wanda aka haɓaka ɗimbin yawa, shine abin da ke samar da ƙwararrun ƙwararrun masu magana da rubutu a yau.

Menene ainihin aikin samfurin harshe yana yi?

Samfurin harshe yana ƙididdige yuwuwar abin da ke zuwa a jere, kuma tsara tana aiki ta maimaita tsinkaya da haɗa alama ta gaba.

Menene maɓalli maɓalli na farkon ƙirar n-gram?

Samfuran N-gram sun dogara da kirga gajerun jerin kalmomi, don haka sun yi amfani da mahallin dogon zango mara kyau kuma sun gaza kan jimlolin da ba su taɓa gani ba.

Me yasa ake kiran horon samfurin harshe 'mai kula da kai'?

Madaidaicin alama na gaba ya riga ya kasance a cikin rubutu, don haka ƙirar ta ƙirƙira maƙasudin sa ba tare da annotation na hannu ba.

Menene ma'anar 'autoregressive' ga samfurin harshe?

Samfuran autoregressive suna haifar da alamar rubutu ta alama, suna daidaita kowane sabon tsinkaya akan duk abin da aka ƙirƙira ya zuwa yanzu.

Wane aikin asara ake amfani da shi don horar da ƙirar harshe?

Horon yana rage girman giciye-entropy, yana ba da lada ga ƙirar don sanya babban yuwuwar zuwa ainihin alamar gaba da aka lura a cikin rubutun horo.