T5 da Canja wurin Rubutu zuwa Rubutu
T5 (Tsarin Canja wurin Rubutu-zuwa-Text), daga Google a cikin 2019, yana sake tsara kowane aikin NLP, fassarar, taƙaitawa, rarrabuwa, har ma da koma baya, azaman ciyarwa a rubutu da samun rubutu.
Dubawa
This single unified format lets one model and one training recipe handle dozens of tasks.
Zurfafa nutsewa
Babban ra'ayin T5 shine cewa kowane aikin harshe za a iya jefa shi azaman rubutu-zuwa-rubutu: shigarwar kirtani ce tare da prefix ɗin ɗawainiya, kuma abin da ake fitarwa koyaushe kirtani ne. Fassara ya zama 'fassara Ingilishi zuwa Jamusanci: ...' yana samar da rubutun Jamusanci; jin ya zama 'sst2 jimla: ...' yana samar da ainihin kalmar 'tabbatacce' ko 'marasa kyau'. Yana amfani da cikakken encoder-decoder Transformer, sabanin BERT mai rikodin-kawai ko mai gyara-kawai GPT. T5 an riga an horar da shi akan C4 corpus (Colossal Clean Crawled Corpus, ~ 750GB na rubutun gidan yanar gizo mai tsabta) tare da maƙasudin cin hanci da rashawa: bazuwar alamun alamun ana rufe su kuma ana maye gurbinsu da alamun sentinel, kuma samfurin ya koyi samar da tazarar da ya ɓace. Binciken da ya biyo baya ya kwatanta tsarin gine-gine, maƙasudai, da girman saitin bayanai don nemo abin da ke canja wuri mafi kyau.
Fahimtar Fasaha
T5's pretraining masks na daɗaɗɗen tazara maimakon alamomi guda ɗaya. Kowane tazarar abin rufe fuska ana maye gurbinsu da wata alama ta musamman a cikin shigarwar, kuma mai yankewa yana samar da saƙon da ke biye da ainihin abinda ke ciki. Wannan ƙin cin hanci da rashawa ya fi inganci fiye da abin rufe fuska ɗaya na BERT. Zane-zane mai rikodin rikodin tare da cikakkiyar kulawar giciye yana barin mai ƙaddamarwa ya halarci gabaɗayan shigar da aka shigar yayin samar da fitarwa ta atomatik.
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 T5 da Canja wurin Rubutu zuwa Rubutu
Tsarin rubutu-zuwa-rubutu ya zama mai tasiri sosai: zuriyar da aka saurara kamar FLAN-T5 gabaɗaya zuwa ayyukan da ba a gani ba daga umarnin harshe na halitta, kuma tsarin haɗe-haɗe yana hasashe manyan samfuran harshe na yau da kullun. Yi tsammanin ci gaba da amfani da na'urori masu rikodin T5 don taƙaitawa, fassara, da tsararrun tsarawa, da bambance-bambancen harsuna da yawa kamar mT5 da magada masu fa'ida mai inganci, kamar yadda ƙirar ƙira-kawai ke mamaye aikace-aikacen taɗi na buɗe ido.
Aiwatar da Gaskiyar Duniya
Takaitacciyar taƙaitawa: prefixing 'takaitawa:' kafin labarin ya sa T5 ta haifar da taƙaitaccen bayani a cikin kalmominsa.
Fassara na'ura: samfurin T5 guda ɗaya yana sarrafa nau'i-nau'i na harshe da yawa ta hanyar prefixes kamar 'fassara Turanci zuwa Faransanci:'.
FLAN-T5 yana bin umarnin yaren yanayi don amsa tambaya da tunani ba tare da takamaiman aikin sake horarwa ba.
Amsa tambaya a rufe-littafi: T5 yana amsa tambayoyin gaskiya kai tsaye kamar rubutun da aka ƙirƙira, yana zana ilimin da aka adana a cikin ma'aunin sa.
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
Ƙayyade tsarin fitarwa, sautin, da ma'auni masu inganci kafin fitowa.
Amsa a ƙasa tare da amintattun tushe a duk lokacin da daidaito ya shafi mahimmanci.
Ajiye wurin binciken ɗan adam don abubuwan da ake samu masu girma.
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 T5 and Text-to-Text Transfer 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
Jagora na gaba
Cire Dangantaka daga Rubutu
Tambayoyin da ake yawan yi
What is T5 and Text-to-Text Transfer?
T5 (Tsarin Canja wurin Rubutu-zuwa-Text), daga Google a cikin 2019, yana sake tsara kowane aikin NLP, fassarar, taƙaitawa, rarrabuwa, har ma da koma baya, azaman ciyarwa a rubutu da samun rubutu. Wannan tsarin haɗin kai guda ɗaya yana ba da damar samfuri ɗaya da girke-girke na horo guda ɗaya su gudanar da ayyuka da yawa.
Menene ma'anar ma'anar T5?
T5 yana jefa fassarar, taƙaitawa, rarrabuwa, da ƙari cikin tsarin rubutu-zuwa-rubutu ɗaya ta amfani da prefixes na ɗawainiya.
Menene Transformer architecture yayi amfani da T5?
Ba kamar BERT mai rikodin-kawai ko mai gyara-kawai GPT ba, T5 yana amfani da cikakkiyar mai canza rikodin rikodi tare da jan hankali.
Wace manufar horarwa ta farko T5 ke amfani da ita?
T5 yana rufe bazuwar tazara na alamomi, yana maye gurbin kowanne da ma'auni, kuma yana horar da na'ura don sake haifar da bacewar tazarar.
Wanne corpus aka riga aka horar da shi T5?
An horar da T5 akan C4, kusan 750GB tsabtace yanki na Rubutun Yanar Gizo na Jama'a.
Ta yaya T5 ke wakiltar aikin rarrabuwa kamar nazarin ji?
Saboda komai na rubutu-zuwa-rubutu, T5 yana fitar da lakabin aji azaman kirtani na kalma maimakon fihirisar ajin lambobi.