Microgpt

MicroGPT ƒe Dzidziɖedzi: Nusita AI ƒe Kpɔɖeŋu Suewo Le Tɔtrɔm le Alesi Asitsahawo Wɔa Dɔe

Le ƒe ʋee siwo va yi me la, AI duɖimekekea nye nusi ŋu wotsia dzi ɖo ɖeka ko: lolo wu. Parameters geɖe wu, hehenana data geɖe wu, akɔntabubu geɖe wu. Gake tɔtrɔ kpata aɖe si me ɖoɖoezizi le le edzi yim le mɔ si to vovo na ema nu. MicroGPT — ʋuʋu yi AI ƒe kpɔɖeŋu siwo le sue, siwo ku ɖe dɔ tɔxɛ aɖe ŋu — le kpe ɖom edzi be suewo ate ŋu anye aɖaŋudzela, kabakaba, eye wòawɔ dɔ wu sã na asitsaha siwo hiã AI ŋutɔŋutɔ be woawɔ dɔ, ke menye be woawɔ dɔ ɖe ame dzi ko o. Togbɔ be mɔ̃ɖaŋudɔwɔƒe gãwo ti nunya si wowɔ le xexeame katã yome tim hã la, asitsaha suewo kple gãwo le ekpɔm be kpɔɖeŋu si le bɔbɔe si wona hehee le woawo ŋutɔ ƒe nyatakakawo dzi ate ŋu awɔ dɔ wu behemoth si ƒe didime nye triliɔn le dɔ siwo le vevie ŋutɔŋutɔ me: asisiwo ƒe biabiawo ŋuɖoɖo, adzɔxegbalẽviwo wɔwɔ, ɖoɖowɔwɔ ɖe ɣeyiɣi siwo woɖo ɖi ŋu, kple dɔwɔwɔwo dzi yiyi AI ƒe gazazã si nye xexlẽdzesi ade manɔmee.

Nuka Tututue Nye MicroGPT?

MicroGPT menye adzɔnu alo dɔwɔƒe ɖeka o — enye aɖaŋuwɔwɔ ƒe xexemenunya. Nya sia ɖɔ AI ƒe kpɔɖeŋu siwo woɖo koŋ le sue, zi geɖe la, woƒe nɔnɔmewo tso miliɔn 1 va ɖo biliɔn 3, siwo ŋu wotrɔ asi le nyuie be woabi ɖe dɔ kpui aɖewo me tsɔ wu be woadze agbagba awɔ nusianu. To vovo na gbegbɔgblɔ gãwo ƒe kpɔɖeŋu siwo wozãna le mɔ gbadza nu (LLMs) siwo hiã GPU ƒuƒoƒo gãwo eye ateŋu axɔ ga si wu $10 le API yɔyɔ 1,000 ɖesiaɖe me na dɔ sesẽwo la, MicroGPT ƒe kpɔɖeŋuwo ateŋu awɔ dɔ le kɔmpiuta si wozãna ɖaa dzi, $5/ɣleti ƒe alilikpo ƒe kpɔɖeŋu, alo gɔ̃ hã wotsɔe de asitsadɔwɔɖoɖo me tẽ.

Nukpɔsusua va xɔ ŋkɔ le ƒe 2023 ƒe nuwuwu esime numekulawo ɖee fia be 1.3-billion-parameter model si woɖɔ ɖo nyuie ɖe domain-specific data dzi ateŋu asɔ kple GPT-3.5 alo awɔ dɔ wu le targeted benchmarks me. Kaka ƒe 2025 naɖo la, habɔbɔ si le ʋuʋu ɖi la wɔ ɖoɖo gbogbo aɖewo — TinyLlama, Phi-3 Mini, Gemma 2B — si na wònɔ bɔbɔe na dɔwɔlawo be woatsɔ micro-scale AI ade dɔwɔwɔ ƒe nɔnɔmewo me. Gbedasi la me kɔ: mèhiã Ferrari ne keke si woɖɔ ɖo nyuie na nèɖo teƒe ma ke kabakaba wu, le ʋuwo ƒe zɔzɔ me o.

Le asitsalawo gome la, tɔtrɔ sia le vevie ŋutɔ. Efia be womegatu AI ɖe dɔwɔƒewo ƒe nubablawo kple nyatakakawo ŋuti dzɔdzɔmeŋutinunya ƒe ƒuƒoƒowo megbe o. Solopreneur si le hehenana dɔwɔƒe aɖe dzi kpɔm alo ame 20 ƒe dɔwɔƒe si kpɔa asisiwo ƒe adzɔxegbalẽviwo dzi ateŋu atsɔ nunya ƒe nuwo wɔwɔ le wo ɖokui si woƒe dɔwɔwɔwo me evɔ mase mɔ̃ ɖeka ƒe nusɔsrɔ̃ ƒe kɔpi gɔme o — nenye be wozã mɔnu nyuitɔ.

Nusitae Massive Models nye Overkill na 90% le Asitsadɔwo ƒe Dɔwɔwɔwo me

Bu asitsaha sue aɖe si bɔ ƒe gbesiagbedɔwo ŋu kpɔ. Ele be aƒetɔa naɖo e-mail siwo akplɔe ɖo ɖe ame siwo kplɔa ame, aɖo kpekpeɖeŋunatike siwo va la ɖe hatsotsowo me, awɔ kwasiɖa sia kwasiɖa ƒe fetu ŋuti nyatakaka kpui aɖe, eye wòaŋlɔ nya aɖe ɖe hadomenyatakakadzraɖoƒewo. Dɔ siawo dometɔ aɖeke mehiã be woabu tamebubu ŋutete si le kpɔɖeŋu si ƒe didime nye biliɔn 175 ŋu o. Le nyateƒe me la, dɔ siawo ɖoɖo ɖe LLM gã aɖe dzi hea ɣeyiɣi didi, nusiwo womate ŋu agblɔ ɖi o, kple gazazã si gblẽa dɔwɔwɔ me vevie vɛ.

Ƒe 2025 ƒe numekuku aɖe si Stanford ƒe Amegbetɔ ƒe AI Dɔwɔƒe wɔ ɖee fia be asitsanyawo ƒe AI zazã ƒe nɔnɔmewo ƒe 87% lɔ dɔ ƒomevi vovovo siwo mede atɔ̃ o ɖe eme, eye be kpɔɖeŋu sue siwo ŋu wotrɔ asi le nyuie la ɖo 94% ƒe nyateƒetoto gbɔ le dɔ siawo me ne wotsɔe sɔ kple 96% na liƒodzi kpɔɖeŋuwo — le gazazã ƒe akpa blaatɔ̃lia me. Ðeko 2% ƒe nyateƒetoto ƒe ŋgɔyiyi suetɔ la meɖo kpe gazazãa dzi na habɔbɔ akpa gãtɔ o. Ne èle aƒexɔfetu wɔm alo le asisiwo ƒe nyaŋuɖoɖowo ɖom ɖe hatsotsowo me la, "te ɖe eŋu nyuie" le $0.001 le biabia ɖesiaɖe me aƒo "perfect" le $0.05.

Esia tututu tae mɔ̃wo abe Mewayz ene ƒo wo ɖokui ɖe modular AI automation me tsɔ wu be woatsɔ bolting ɖe monolithic AI layer ɖeka dzi. Esi wònye be modules tɔxɛ 207 siwo ƒo nu tso CRM, invoicing, HR, payroll, fleet management, kple analytics ŋu ta la, Mewayz zãa automation si me nunya le le afisi tututu wòhiã le — le module ɖesiaɖe ƒe nɔnɔme tɔxɛ me — tsɔ wu be wòawɔ funneling asitsatsa ƒe ɖoɖo ɖesiaɖe to AI engine si ƒe lolome sɔ na amewo katã dzi.

Asitsatsa ƒe Viɖe Atɔ̃ Siwo Le Going Micro

Tɔtrɔ yi MicroGPT gɔmeɖosewo gbɔ naa viɖe ŋutɔŋutɔ siwo woate ŋu adzidze siwo le vevie na asitsahabɔbɔ siwo tsɔa ɖe le margin me wu kpɔɖeŋu xɔtuɖaŋuwo. Viɖe siawo gasẽna ɖe edzi ne asitsahabɔbɔwo le dzidzim ɖe edzi, si wɔnɛ be mɔnu suea va le dzedzem ɖe edzi le ɣeyiɣi aɖe megbe.

ƒe nyawo
• Gazazã dzi ɖeɖe kpɔtɔ 80-95%: 2B-parameter ƒe kpɔɖeŋu si woɖɔ ɖo nyuie ƒe duƒuƒu le alilikpo ƒe kpɔɖeŋu si mebɔ o dzi xɔa $15 va ɖo $50 ɣleti sia ɣleti. API yɔyɔ siwo sɔ na liƒo ƒe kpɔɖeŋu na dɔwɔwɔ ƒe agbɔsɔsɔme ɖeka ateŋu agbɔ $500 ɣleti sia ɣleti bɔbɔe.
• Latency under 200ms: Micro models wɔa dɔ le milisekɔnd 50-200 me ne wotsɔe sɔ kple sɛkɛnd 1-5 na model gã siwo woxɔ ɖe alilikpo me. Le dɔwɔɖoɖo siwo dze ŋgɔ asisiwo abe chatbots alo didi le ɣeyiɣi ŋutɔŋutɔ me gome la, vovototo sia ɖea zãla ƒe nuteƒekpɔkpɔ gɔme.
• Nyatakakawo ƒe adzamenyawo le gɔmedzedzea me: Ne kpɔɖeŋua le dɔ wɔm le wò xɔtuɖoɖowo dzi — alo le SaaS mɔ̃ si nètia me — asisiwo ƒe nyatakakawo medzona le wò nutoa me gbeɖe o. Le asitsaha siwo le lãmesẽnyawo gbɔ kpɔkpɔ, senyawo, alo ganyawo gbɔ kpɔkpɔ me gome la, esia menye atsyɔ̃ɖoɖo o; enye sedziwɔwɔ ƒe nudidi.
• Nuwɔna si woate ŋu agblɔ ɖi: Kpɔɖeŋu sue siwo ŋu wotrɔ asi le nyuie ɖe nyatakakadzraɖoƒe tɔxɛwo dzi la wɔa emetsonu siwo mewɔa tɔtrɔ o, siwo woate ŋu agbugbɔ awɔ. Afɔku aɖeke mele eme be AI la nahe ŋuɖoɖo si to vovo na adzɔnu ƒe ŋkɔ vɛ zi ɖeka alo awɔ susumenuwo le ɖoɖo aɖe si meli o ŋu o.
• Offline ŋutete: Edge-deployed micro models wɔa dɔ internet kadodo manɔmee, si wɔnɛ be agbledelawo, adzɔge dɔwɔlawo, kple asitelefon-gbãtɔ dɔwɔƒewo te ŋu kpɔa AI-ŋusẽ ƒe nɔnɔmewo le afisiafi.

ƒe nyawo

Viɖe siawo menye nukpɔsusuwo o. Aƒewo dzikpɔha aɖe si le Melbourne ka nya ta be yewoɖe yewoƒe aƒehayalawo ƒe kadodo ŋuti gazazãwo dzi kpɔtɔ 73% le esi yewotrɔ tso AI chatbot si wozãna le mɔ gbadza nu dzi va zu micro model si wona hehee le yewoƒe hayahaya ƒe nubablawo, beléle na wo ƒe ɖoɖowo, kple FAQ nyatakakadzraɖoƒe ɖeɖeko ŋu vɔ megbe. Kpɔɖeŋu suetɔ ɖo aƒehayalawo ƒe biabiawo ŋu nyuie wu elabena nusi wòle be wòanya koe wònya.

Alesi MicroGPT Gɔmeɖosewo Doa Ŋusẽ Ðe Egbegbe Asitsahabɔbɔwo Ŋu

Asitsatsa ƒe kɔmpiuta dɔwɔɖoɖo dɔwɔƒe siwo dze aɖaŋu wu la tsɔ MicroGPT xexemenunya de wo ɖokui me evɔ womedo bobloe kokoko o. Le esi teƒe be woado ŋugbe be "AI-powered everything" la, wotsɔa nunya si wotu ɖe taɖodzinu aɖe ta dea dɔwɔwɔ ɖekaɖekawo me. CRM module zãa classifier sue aɖe tsɔ dea dzesi leads. Invoicing module zãa pattern-recognition model tsɔ dea dzesi nusiwo mesɔ o. Ðoɖowɔɖi ƒe mɔ̃ zãa asitɔtrɔ ƒe mɔnu si le bɔbɔe tsɔ ɖea agbalẽŋɔŋlɔ ƒe masɔmasɔwo dzi kpɔtɔna.

Modular mɔnu sia ɖe alesi Mewayz subɔa eƒe zãla 138,000+ le dɔwɔƒewo katã me fia. Le esi teƒe be woatsɔ AI ƒe kpeɖeŋutɔ ɖeka aɖo nuƒolanɔƒe bliboa dzi la, Mewayz ƒe module 207 ɖesiaɖe tsɔa automation si wowɔ ɖe eƒe dɔ tɔxɛa nu la dea eme. Fetu ƒe ɖoɖoa sea adzɔxexe ƒe akɔntabubuwo kple se siwo dzi woawɔ ɖo gɔme. Agbalẽŋɔŋlɔ ƒe ɖoɖoa naa ɣeyiɣi siwo woatsɔ awɔ ɖoɖoe la nyona ɖe edzi le ŋutinya me nɔnɔmewo nu. CRM la auto-segments kadodowo to nuwɔna ŋuti nyatakakawo zazã me. Module ɖeka aɖeke medzea agbagba be yeanye nusianu o — eye mɔxexeɖedɔléle ma tututue nana kakaɖedzi le ɖoɖoa ŋu.

ƒe nyawo

"Asitsatsa ƒe AI ƒe etsɔme menye ahɔhɔ̃ gã ɖeka o — enye nunya sue tɔxɛwo ƒe kadodo, siwo dometɔ ɖesiaɖe bi ɖe domenyinyi ɖeka me. Dɔwɔƒe siwo se esia gɔme la awɔ dɔ nyuie wu amesiwo gakpɔtɔ le akunyawɔwɔ, nusianu wɔwɔ ƒe AI lalam be wòakpɔ woƒe kuxiwo katã gbɔ zi ɖeka."

ƒe nyawo

Xɔtuɖaŋu ƒe tiatia sia hã fia be asitsalawo ateŋu axɔ AI vivivi. Freelancer ateŋu adze egɔme kple Mewayz ƒe link-in-bio dɔwɔnu kple invoicing module le free plan la dzi, emegbe vivivi awɔ CRM, analytics, kple HR modules ŋudɔ ne woƒe ƒuƒoƒoa le tsitsim — wo dometɔ ɖesiaɖe atsɔ eya ŋutɔ ƒe layer of intelligent automation vɛ evɔ mahiã be woawɔ wholesale platform migration alo AI strategy overhaul.

Xexeme Ŋutɔŋutɔ ƒe Dɔwɔwɔwo: Afisi Micro Models Wɔ dɔ Wu Giants

Kpeɖodziawo le dzidzim ɖe edzi le dɔwɔƒewo katã me. Le e-asitsatsa me la, asitsala siwo zãa adzɔnuwo ƒe kafukafu ƒe kpɔɖeŋu sue siwo ŋu wotrɔ asi le nyuie le woawo ŋutɔ ƒe agbalẽdzraɖoƒe ŋuti nyatakakawo gblɔ be amewo ƒe asiƒoƒo dzi ɖe edzi 22% wu amesiwo zãa kafukafu API siwo wozãna le mɔ vovovowo nu. Le lãmesẽnyawo gbɔ kpɔkpɔ me la, atikewɔƒe siwo zãa micro models hena appointment triage ɖe no-show rates dzi kpɔtɔ 31% to nyagblɔɖi kple kadodo do ŋgɔ na dɔnɔ siwo le afɔku me. Le dɔnyala ƒe dɔwɔƒewo la, senyawo gbɔ kpɔƒe siwo zãa nuŋlɔɖiwo ƒe hatsotsowo me toto ƒe kpɔɖeŋu siwo le sue la wɔa nubablawo ŋudɔ kabakaba zi gbɔ zi 4 wu esiwo ɖoa ŋu ɖe nuŋlɔɖi AI si wozãna le taɖodzinu bliboa me ŋu.

Kpɔɖeŋua sɔ: ne kpɔɖeŋua menya nu boo aɖeke tso xexeame ŋu o gake menya nu geɖe tso wò dɔwɔƒea ŋu o la, ewɔa dɔ nyuie wu le wò asitsadɔwo me. Nuɖuɖudzraƒe mehiã AI si ate ŋu aŋlɔ hakpanyawo ahakpɔ akɔntabubu ƒe kuxiwo gbɔ o. Ehiã ɖeka si ateŋu agblɔ Kuɖagbe fiẽnuɖuɖu ƒe didi ɖi le yame ƒe nɔnɔme, nutoa me nudzɔdzɔwo, kple ŋutinya me nutsyɔnuwo nu — eye 50-milion-parameter model ateŋu awɔ ema nyuie.

Etɔxɛ sia ɖea afɔku si me afɔku le wu le asitsatsa ƒe AI me hã dzi kpɔtɔna: susumenuwo kpɔkpɔ. Micro model si wona hehee le wò adzɔnuwo ƒe agbalẽdzraɖoƒe mate ŋu ato adzɔnu si meli o vɛ o. Micro model si wona hehee le wò HR ɖoɖowo ŋu mate ŋu awɔ viɖe aɖe si mèna o. Sidzedze ƒe gɔmeɖoanyi si ŋu woxe mɔ ɖo la zua nusi ɖe dzesi, ke menye seɖoƒe o — eye le asitsaha siwo me kakaɖedzi kple nyateƒetoto nye nusi ŋu womate ŋu awɔ ɖoɖo ɖo o la, viɖe sia nyea nyametsotso.

Wò MicroGPT Aɖaŋuɖoɖo Tutu Kpeɖodzinya Ŋɔŋlɔ Manɔmee

Mɔ si wɔa dɔ wu na asitsaha akpa gãtɔ menye hehenana mɔ̃ siwo wowɔ ɖe ɖoɖo nu tso gɔmedzedzea me o — ke boŋ tiatiawɔƒe siwo wɔ dɔ sesẽ si nye be woatsɔ nunya tɔxɛwo ade dɔwɔnu siwo woate ŋu azã me xoxo. Tu-x-nu-ƒle ƒe sɔsɔŋusẽa doa dzidzɔ na nuƒle wu ne wò dɔ vevitɔ menye AI ŋgɔyiyi o.

Dze egɔme kple wò dɔwɔwɔwo me dzodzro hena dɔ siwo gbugbɔna wɔna, siwo wotu ɖe se dzi siwo xɔa amegbetɔ ƒe gaƒoƒo fifia: nyatakakawo ŋɔŋlɔ, kadodo siwo kplɔe ɖo, nyatakakawo wɔwɔ, ɖoɖowɔwɔ, kple hatsotsowo me toto. Esiawoe nye dɔwɔwɔ ƒe ɖoɖo siwo me micro-scale AI naa ROI enumake. Eyome, da mɔ̃wo kpɔ le nenye be woƒe AI nye modular vavã — si fia be àteŋu awɔ nunya ƒe nɔnɔmewo ŋudɔ le nuto ɖeka me evɔ màxe mɔ ɖe nusianu alo naneke AI ƒe ɖokuitsɔtsɔna me o.

| Nya vevi si wòle be woabia platform dzrala ɖesiaɖe menye "Ðe AI le asiwòa?" — enye "Ðe wò AI se nye dɔwɔwɔ ƒe ɖoɖo tɔxɛ gɔmea, alo ɖe wònye generic layer bolted on top?"

Nyatae: Ele be Nunya Nanye Nusi Womate Ŋu Akpɔ O

MicroGPT habɔbɔa fiaa nusɔsrɔ̃ aɖe si keke ta yi ŋgɔ sasasã wu nunya wɔwɔe: mɔ̃ɖaŋununya nyuitɔ kekeake buna ɖe dɔwɔwɔ me. Mele be ezãlawo nabu model ƒe lolomewo, parameter ƒe xexlẽme, alo inference costs ŋu o. Ele be woaʋu woƒe asitsamɔnu, eye ɖeko nuwo nawɔ dɔ — kabakaba, aɖaŋudzedze, eye wòade pɛpɛpɛ wu alesi wowɔe etsɔ.

Le asitsaha 138,000+ siwo le dɔ wɔm xoxo le Mewayz gome la, esiae nye gbesiagbenudzɔdzɔ. Wodea dzesi gaxɔgbalẽviwo hafi vodadawo ɖoa asisiwo gbɔ. Woxɔa dzesi kplɔlawo hafi nudzralawo ƒe ƒuƒoƒowo ʋua woƒe dashboardwo. Ðoɖowɔɖiwo va nyona ɖe edzi hafi masɔmasɔwo doa mo ɖa. Wo dometɔ aɖeke mebia tso ezãla si be wòase nusi MicroGPT fia gɔme o — eye nya la tututue nye ema. AI nyuitɔ kekeakee nye AI si ŋu mehiã be nàbu gbeɖe o.

| Micro revolution la mele vava ge o — ele afisia xoxo, wotsɔe de dɔwɔnu siwo asitsaha siwo me nunya le zãna gbesiagbe la me kpoo.

ƒe nyawo

Nyabiase Siwo Wobiana Enuenu

Nuka tututue nye MicroGPT?

MicroGPT fia AI ƒe kpɔɖeŋu sue siwo le etɔxɛ ŋutɔ siwo wowɔ na dɔ tɔxɛwo wɔwɔ, si to vovo na taɖodzinu gã siwo wozãna le mɔ gbadza nu abe GPT-4 ene. Wona hehe mɔ̃ siawo siwo le sue be woabi ɖe dɔ kpui aɖewo abe asisiwo subɔsubɔ, nyatakakawo me dzodzro, alo nyatakakawo wɔwɔ me ene. Woƒe susu sia nana be wowɔa dɔ nyuie wu, woƒe asi bɔbɔ, eye wònɔa bɔbɔe be woatsɔ wo ade asitsatsa ƒe dɔwɔwɔ me evɔ gazazã gã kple mɔnu gãwo ƒe sesẽ manɔmee.

Nukatae AI ƒe kpɔɖeŋu suewo nyo na dɔwɔƒewo?

Kpɔɖeŋu suewo abe MicroGPT ene naa viɖe gãwo na dɔwɔƒewo, vevietɔ SMBwo. Woƒua du kabakaba wu, womexɔa ga geɖe o, eye womehiãa akɔntabubuŋusẽ boo o, si wɔnɛ be dɔwɔwɔ ƒe gazazãwo dzi ɖena kpɔtɔna. Woƒe nɔnɔme tɔxɛ fia be zi geɖe la, wowɔa dɔ nyuie wu le woƒe dɔ tɔxɛwo wɔwɔ me eye vodada ʋɛ aɖewo koe mewɔna o. Mɔ̃wo abe Mewayz, kple eƒe modules 207 siwo ƒe home nye $19/mo, ɖe alesi AI dɔwɔnu siawo siwo ŋu wotsia dzi ɖo la va zu nusi ŋu woate ŋu aɖo eye woate ŋu awɔ dɔe na gbesiagbe asitsatsa ƒe dɔwɔnawo fiana.

Aleke mawɔ atsɔ MicroGPT ƒe kpɔɖeŋu awɔ ɖeka kple nye ɖoɖo siwo li fifia?

Zi geɖe la, ƒoƒo ɖekae nyea nusi le tẽ. Wotsɔa MicroGPT ƒe kpɔɖeŋu akpa gãtɔ naa to APIwo dzi alo abe modules siwo wotu do ŋgɔ siwo wowɔ na mɔ̃ tɔxɛwo ene. Le kpɔɖeŋu me, àte ŋu azã dɔwɔƒe aɖe abe Mewayz ene atsɔ AI ƒe ŋutetewo abe seselelãmewo me dzodzro alo ŋuɖoɖo siwo wowɔna le wo ɖokui si ene akpe ɖe wò CRM, kpekpeɖeŋunaƒe, alo asitsatsa dɔwɔnuwo ŋu tẽe evɔ màhiã AI eŋutinunyalawo ƒe ƒuƒoƒo o. Plug-and-play mɔnu sia ɖea mɔ̃ɖaŋununya ƒe mɔxenu si xea mɔ na gege ɖe eme dzi kpɔtɔna ŋutɔ.

Ðe kpɔɖeŋu sue aɖe ƒe ŋutete mede esi lolo abe GPT-4 ene oa?

Menye ɖe eƒe taɖodzinu si woɖo tae o. Togbɔ be MicroGPT ƒe kpɔɖeŋu makpɔ sidzedze gbadzaa, si le xexeame katã le kpɔɖeŋu gã aɖe ŋu o hã la, zi geɖe la, edea pɛpɛpɛ wu eye kakaɖedzi le eŋu na dɔ tɔxɛ si ta wotue ɖo. Bu eŋu be enye aɖaŋudɔwɔla bibi aɖe tsɔ wu aɖaŋudɔwɔla bibi aɖe. Le asitsatsa ƒe dɔwɔnu siwo woɖo taɖodzinu na gome—afisi pɛpɛpɛ, duƒuƒu, kple gazazã le vevie wu dzeɖoɖo ƒe didime—zi geɖe la, kpɔɖeŋu sue tɔxɛ aɖee nyea tiatia si nyo wu.