ROUGE da BLEU Ma'aunin kimantawa
ROUGE da BLEU sune ma'auni na atomatik na aiki don kwatanta rubutun da aka samar da na'ura akan nassoshi na ɗan adam.
Dubawa
BLEU was built for translation and leans on precision; ROUGE was built for summarization and leans on recall.
Zurfafa nutsewa
Duk ma'auni guda biyu suna auna haɗin n-gram tsakanin rubutun ɗan takara da rubutun magana ɗaya ko fiye, amma suna jaddada kwatance daban-daban. BLEU (Ƙaramar Nazarin Harsuna Biyu) tana ƙididdige madaidaicin n-gram (yawanci 1- zuwa 4-grams), yana ninka su ta hanyar geometric, kuma yana zartar da hukuncin taƙaitaccen hukunci don haka tsarin ba zai iya yin makin ta hanyar samar da ɗan gajeren fitarwa ba. ROUGE (Ƙaramar Tunawa-Mai Gabatarwa don Ƙimar Gisting) a maimakon haka ya fi son tunawa: ROUGE-N yana ƙidaya n-grams masu rufewa, ROUGE-L yana amfani da mafi tsayi na gama-gari don ba da lada a cikin tsari ba tare da buƙatar haɗin kai ba. BLEU yayi tambaya 'nawa na abin da tsarin ya ce daidai ne?' yayin da ROUGE ke tambaya 'nawa ne tsarin ya kama?'. Dukansu suna da arha kuma ana iya sake su amma kawai ga kalma ta sama, ta ɓace da ma'ana.
Fahimtar Fasaha
Shirye-shiryen gyaran gyare-gyare na BLEU kowane ɗan takara n-gram yana ƙidayar ƙidayarsa a cikin kowane tunani, yana hana maimaita wasan caca; hukunce-hukuncen takaitawa yana farawa lokacin da fitarwa ya fi guntu fiye da abin da aka ambata. ROUGE-L mafi dadewa-na gaba-gaba yana ɗaukar tsarin matakin jumla da tsarin kalma yayin da yake ba da damar giɓi, kuma ROUGE yakan bayar da rahoton F1 yana haɗa daidaici da tunawa.
Dabarun Tasiri
Kudin da kasafin kuɗi
Hukunce-hukuncen gine-gine suna haifar da aiki da tsadar aiki na shekaru.
Shawarwari masu haske
Ilimin fasaha yana taimaka wa ƙungiyoyi su zaɓi tari mai kyau, ba kawai sabon abu ba.
Kula da inganci
Zaɓuɓɓukan injiniya mafi kyau suna rage abin dogaro a cikin samarwa.
Makomar ROUGE da ma'aunin kimantawa na BLEU
Saboda ma'aunin n-gram suna ba da madaidaicin matches na kalmomi, suna ƙasƙantar da ingantattun fassarori da sake rubutawa da kyau, matsala mai girma yayin da fitowar LLM ta bambanta da nassoshi. Ƙirƙirar ma'auni na tushen kamar BERTScore da koyo awo kamar BLEURT da COMET, da LLM-as-alkali kimantawa, ƙara kari ko maye gurbinsu. Har yanzu, ROUGE da BLEU suna dagewa da sauri, madaidaicin tushe da aka ruwaito a kusan kowace takarda.
Aiwatar da Gaskiyar Duniya
Masu binciken fassarar inji sun ba da rahoton maki BLEU akan ma'auni na WMT don kwatanta ingancin tsarin
Takardun taƙaitawa suna ba da rahoton ROUGE-1, ROUGE-2, da ROUGE-L akan bayanan CNN/DailyMail
Ƙungiyar injiniya tana bin BLEU a cikin CI don gano koma baya lokacin daidaita tsarin fassarar
Samfurin taƙaitawa yana amfani da ROUGE-L azaman rajistan atomatik mai arha kafin gudanar da ƙimar ƙimar ɗan adam
Hatsari & Tsare-tsare
Haɓaka ma'auni ɗaya na iya ɓoye manyan raunin tsarin.
Sau da yawa ana raina kayan more rayuwa da kuma kuɗin kulawa.
Tsaro da gibin lura na iya girma yayin da tsarin ke ƙara haɓaka.
Taswirar Hanya
Ƙayyade latency, inganci, da maƙasudin farashi kafin aiwatarwa.
Alamar ma'auni a ƙarƙashin ainihin kaya da yanayin bayanai.
Kula da kayan aiki don kurakurai, ɗigo, da tasirin mai amfani.
Shirya bijirowa da hanyoyin mayar da martani kafin sikeli.
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 ROUGE and BLEU Evaluation Metrics 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
Ƙimar Layi na Layi da Daskararrun Ƙimar Siffar
Tambayoyin da ake yawan yi
What is ROUGE and BLEU Evaluation Metrics?
ROUGE da BLEU sune ma'auni na atomatik na aiki don kwatanta rubutun da aka samar da na'ura akan nassoshi na ɗan adam. An gina BLEU don fassarar kuma yana dogara akan daidaito; ROUGE an gina shi don taƙaitawa kuma ya dogara akan tunawa.
Wanne awo ne aka ƙera asali don fassarar inji kuma ya jaddada daidaito?
An ƙirƙiri BLEU don fassara kuma an dogara ne akan ingantaccen n-gram wanda aka gyara tare da taƙaitaccen hukunci.
Menene ROUGE ya fi karkata zuwa ga?
ROUGE yana tsaye ne don Ƙwararrun Tunawa-Mai Gabatarwa don Ƙimar Gisting kuma yana jaddada yawan abubuwan da aka kama.
Menene taƙaitaccen hukunci na BLEU ya hana?
Hukuncin taƙaitawa yana saukar da BLEU lokacin da ɗan takarar ya fi guntu fiye da tunani, dakatar da tsarin daga haɓaka daidaito tare da fitarwa mai ƙarfi.
Menene ma'aunin ROUGE-L?
ROUGE-L yana amfani da mafi dadewa na gama gari, yana ba da lada cikin tsari yayin ba da damar gibi.
Menene maɓalli na maɓalli na haɗin gwiwa na BLEU da ROUGE?
Dukansu sun dogara da madaidaicin madaidaicin kalma/n-gram, don haka suna ƙasƙantar da ingantattun fassarori waɗanda suka bambanta da ƙamus.