Ungayisebenzisa kanjani i-Deepseek R1 ku-Perplexity

Umhlaba wobuhlakani bokwenziwa uphenduke wabheka phansi ngokufika kwemodeli I-DeepSeek R1, uguquko olungagcinanga ngokushiya ochwepheshe nezinkampani bengawuvali umlomo, kodwa futhi ludale impikiswano nokusola okuthile emkhakheni wezobuchwepheshe. Ukuhlanganiswa kwe-DeepSeek R1 ku-Perplexity AI nezinye izinkundla kuphawule inguquko. kulabo abafuna ukuthola okuningi kumathuluzi e-Artificial Intelligence, kokubili kubasebenzisi ngabanye namabhizinisi.

Lesi sihloko sichaza konke odinga ukukwazi mayelana ne-DeepSeek R1 ku-Perplexity: kusukela endleleni esebenza ngayo nalokho ekunikezayo ekusebenzeni, kuya ezinhlelweni zayo ngaphezu kwamanye amamodeli, izici ezibalulekile mayelana nokuhlolwa nokuphepha, kanye nokuqhathanisa nezinye izixazululo emakethe.

Iyini i-DeepSeek R1 futhi kungani iguqule ubuhlakani bokwenziwa?

 

DeepSeek R1 es imodeli ye-AI ethuthukisiwe eyenziwe yinkampani yaseShayina iHigh-Flyer, eholwa ngu-Liang Wenfeng. Kuyimodeli edizayinelwe imisebenzi yokucabanga ethuthukisiwe, ikakhulukazi ukwenza kahle kwizibalo, ukuhlela, ukuhlaziya okunengqondo, nolimi lwemvelo. Ingxubevange Yochwepheshe (i-MoE) yezakhiwo Ngamapharamitha ayizigidi eziyizinkulungwane ezingama-671 nokuqeqeshwa okuqinisiwe, kusezingeni elifanayo - futhi ezimweni eziningi ngenhla - amanye ama-AI asezingeni eliphezulu njenge I-OpenAI, I-Google Gemini y I-Microsoft Copilot.

Elinye lamaphuzu alo aqinile ukusebenza kahle, njengoba noma ubukhulu bayo bukhulu kakhulu, yenza kuphela ingxenye yemingcele yayo embuzweni ngamunye, okuvumela isivinini kanye nezindleko eziphansi kakhulu uma kuqhathaniswa nezimbangi ngqo. Isibonelo, i-DeepSeek R1 ingabiza izikhathi ezifika kwezingu-27 ngaphansi kwamathokheni ayisigidi uma kuqhathaniswa namamodeli afana ne-OpenAI's o1.

Mayelana nokubuyekezwa kanye nezilinganiso, i-DeepSeek R1 yamukelwe ngentshiseko izinkampani ezifana ne-Nvidia ne-OpenAI, ezibone amandla ayo kanye negalelo layo ekuthuthukisweni kwe-AI emisebenzini eyinkimbinkimbi.

Ukuhlanganiswa kwe-DeepSeek R1 ne-Perplexity AI: ukugxumela phambili

 

I-Perplexity AI, enye yezinjini zokusesha ezisekelwe ku-AI namuhla, ithathe isinyathelo Hlanganisa i-DeepSeek R1 kuwebhu yakho nasezinsizeni zohlelo lokusebenzaLokhu kusho ukuthi abasebenzisi bangakwazi ukufinyelela amandla e-DeepSeek ngokuqondile ku-Perplexity interface, besizakala ngamakhono ayo okucabanga kanye nokusebenziseka kalula.

Ukuqaliswa kwe-DeepSeek R1 e-Perplexity kugqama ngezizathu ezimbalwa: aspectos clave:

• Imiphumela engahloliwe: Ukuxakeka kuguqule imodeli ukuze kususwe imikhawulo yasekuqaleni ebekwe uhulumeni waseShayina ezihlokweni zezepolitiki noma ezibucayi, okuvumela inkululeko enkulu yokubonisana nokubeka obala kolwazi.
• Ubumfihlo kanye nesitoreji esivikelekile: Idatha ecutshungulwe yi-DeepSeek R1 e-Perplexity igcinwa kumaseva e-United States naseYurophu, iqeda ukukhathazeka ngokulawula noma ukufinyelela okuvela e-China.
• Umthombo ovulekile nokutholakala: I-DeepSeek R1 iyimodeli yomthombo ovulekile, okwenza kube lula ukuhlanganisa nokusebenzisa yibo bobabili abathuthukisi namabhizinisi.

Ungayisebenzisa kanjani i-DeepSeek R1 ku-Perplexity: izinyathelo nezincomo

Isinyathelo sokuqala Finyelela inkundla ye-Perplexity, etholakala kokubili kuwebhu nakuhlelo lokusebenza lwe-iOSUma ungakabi nayo i-akhawunti, dala eyodwa kalula. Uma usungenile, landela lezi zinyathelo ukuze usebenzise i-DeepSeek R1:

1. Thola ibhokisi losesho bese ukhetha inketho ethi “Ukubonisana nge-R1” (kungase kubonakale njengokuthi “ukubonisana” nje kwezinye izinguqulo).
2. Ungakhetha ukwenza amafonti ngendlela oyifisayo lapho ungathola khona ulwazi, okwengeza ukulawula okwengeziwe kwemiphumela.
3. Faka umbuzo wakho futhi ivumele i-DeepSeek R1 isebenzise wonke amakhono ayo okucabanga. Uma udinga ukubuka imithombo yolwazi, kuhlala kunethebhu ezinikele ebizwa ngokuthi "Imithombo."
4. Generación de imágenes: Isici esiyingqayizivele sikuvumela ukuthi ukhiqize izithombe ngesithonjana esihambisanayo, sisenze siguquguquke nakakhulu kumaphrezentheshini noma imibhalo ebonakalayo.

Okubalulekile: Kubasebenzisi bamahhala, ukufinyelela ku-DeepSeek R1 kuvame ukukhawulelwa emibuzweni emi-3-5 ngosuku. Kodwa-ke, labo abakhetha ukubhalisa kwe-Perplexity Pro bangenza ukusesha okungafika ku-500 ngosuku, okuyenza ilungele abacwaningi, abafundi, noma izicelo zebhizinisi ezijulile.

Izici ezihlukanisayo uma ziqhathaniswa namanye amamodeli we-AI

• Coste y eficiencia: I-DeepSeek R1 ishibhile kakhulu ukuyisebenzisa kunamanye ama-AI asezingeni elifanayo, enikeza inani elingenakuhlulwa ngamathokheni ayisigidi.
• Izimpendulo ezisobala nezitholakalayo: Enye yezinzuzo ezinkulu zokusebenzisa i-DeepSeek R1 ku-Perplexity ukuthi ibonisa izinyathelo ezimaphakathi zokucabanga kwayo futhi icaphuna imithombo esetshenzisiwe, into engavamile emqhudelwaneni.
• Usekelo lombuzo omude: Ingakwazi ukuphatha okuqukethwe okungafika kumathokheni angu-128.000, okukuvumela ukuthi uhlukanise amadokhumenti amade noma izintambo zombhalo eziyinkimbinkimbi.
• Okukhiphayo okungahloliwe: Inguqulo ye-Perplexity R1 1776 iqeqeshelwe ukuthi ingavimbeli izimpendulo ezihlokweni ezibucayi, ngokungafani ne-DeepSeek R1 yasekuqaleni, engase yenqabe ukubhekana nezinkinga ezithile.
• Integraciones avanzadas: Ikuvumela ukuthi uhlanganyele nezinhlelo zokusebenza eziningi, izindawo zokudlela, ikhodi yokususa iphutha, noma uhlele izikhumbuzi zonke kusuka kusixhumi esibonakalayo esifanayo, uqondise kokubili umsebenzi wakho siqu kanye nochwepheshe.

Imodeli ye-Perplexity R1 1776: i-DeepSeek engenazo izihlungi

Enye yentuthuko ephawuleka kakhulu kube ukudalwa kwe okuhlukile kwe-R1 1776, kuphela i-Perplexity, esusa ngokuphelele ukucwaninga nemikhawulo ebekwe kumodeli yasekuqaleni ye-DeepSeek R1. Lo "R1 1776" uqeqeshwe ngemiyalezo engaphezu kuka-40.000 enezilimi eziningi kanye nezihloko ezibucayi ezingu-300 ukuze kuqinisekiswe izimpendulo ezivulekile emibuzweni yomlando, yezombusazwe, kanye namalungelo abantu. Inombolo "1776" ibhekisela ngokomfanekiso enkululekweni, ihlehlela emuva onyakeni we-American Declaration of Independence.

Ngenxa yalo mzamo, abasebenzisi manje bangakwazi Thola ulwazi olunembile nokucabanga ngaphandle kokuthi imodeli yenqabe ukukhuluma ngezihloko ezazinqatshelwe ngaphambiliniUcwaningo lwangaphakathi olwenziwe yi-Perplexity kanye nemidiya ekhethekile lukhombisa ukuthi imibuzo engaphezu kuka-1000 ezihlokweni ezazivinjwe ngaphambilini iphendulwe ngempumelelo, ngezinga lokunemba elingaphezu kakhulu kwamanye ama-AI okuthiwa "ayihloliwe".

Izinzuzo zamabhizinisi nabasebenzisi abathuthukile

Ukutholakala kwemodeli ye-DeepSeek R1 ezisekelweni ezisezingeni eliphezulu njenge-Azure AI Foundry, GitHub, kanye ne-Google Vertex AI yenze yaba lula. ukutholwa yizinkampani ezinkuluEqinisweni, iMicrosoft kanye ne-Google bakuqinisekisile ukuhlanganiswa kwe-R1 phakathi kwamamodeli abo atholakalayo, beyingeza kukhathalogi yamamodeli angaphezu kuka-1.800 avulekile nathengiswayo.

Izinzuzo eziphawuleka kakhulu zebhizinisi nendawo yobungcweti yilezi:

• Ukukala nesivinini: I-DeepSeek R1 ingahlanganiswa ngokushesha nangokuvikelekile ekugelezeni komsebenzi webhizinisi kusetshenziswa ingqalasizinda ethembekile efana naleyo evela ku-Microsoft ne-Google.
• Ukuhlola ukuphepha okuphelele: I-Microsoft, isibonelo, ibeke imodeli ekuhlolweni kweqembu elibomvu kanye nokubuyekezwa kokuphepha ukuze kuncishiswe noma yiziphi izingcuphe ezingaba khona.
• Flexibilidad de integración: Ingatshalwa kalula kusuka kumakhosombe afana ne-Hugging Face, okwenza kube lula ukufinyelela ngisho nakubathuthukisi abazimele.

Izinhlelo zentengo kanye nokufinyelela ku-DeepSeek R1 ku-Perplexity

I-DeepSeek i-R1 Itholakalela zonke izinhlobo zabasebenzisi, kodwa umuzwa uyahlukahluka kuye ngohlelo olukhethiweKulabo abafuna ukuhlola ngokushesha noma ukuhlanganisa amakhono ayo, inguqulo yamahhala inikeza imibuzo emi-3-5 ngosuku.

• Inguqulo yamahhala: Ivumela ukusesha okungafika ku-3-5 ngosuku kusetshenziswa imodeli ye-R1 ku-Perplexity. Kwanele kulabo abadinga kuphela imibuzo ethile.
• I-Perplexity Pro: Ngo-$20 ngenyanga (futhi mahhala kumakhasimende o-opharetha abathile njenge-Movistar unyaka), ububanzi bunwebeka bufinyelele ku-500 imibuzo yansuku zonke nezici ezikhethekile, ezilungele labo abadinga ukusetshenziswa okujulile noma okuphrofeshinali.

Aprovechar la I-Perplexity Pro umtitilizo wamahhala wonyaka Kungathakazelisa ikakhulukazi uma uyikhasimende labasebenzisi abafinyelele izivumelwano nenkundla, into eseyaduma eSpain naseLatin America.

Umbono Wesikhathi Esizayo: Ingabe i-DeepSeek i-R1 iyizinga Elisha Ekucabangeni kwe-AI?

Ukuvela kwe-DeepSeek R1 futhi, ngaphezu kwakho konke, Ukufika kwakhe ngokunganaki ePerplexity kuyishintshile imithetho yomdlalo.Ngenxa yokusebenzisana kahle kwezindleko, amandla okucabanga, kanye nenkululeko yetimu, kulindeleke ukuthi ibe ngenye yamamodeli asezingeni elifanele eminyakeni ezayo, ekhaya nasemsebenzini nasezilungiselelweni zemfundo.

La tendencia indica que Izinkundla ezengeziwe zizohlanganisa amamodeli avulekile njenge-DeepSeek R1, futhi umsebenzisi uzothola inkululeko, obala nokulawula ulwazi.Noma kunjalo, impikiswano mayelana nokuvikeleka nokuqukethwe okubucayi izoqhubeka, ngakho sizobona intuthuko eqhubekayo ukuze sifinyelele ibhalansi phakathi kokusungula izinto ezintsha nokuzibophezela.

Kuyacaca ukuthi sisesigabeni esijabulisayo emlandweni wobuhlakani bokwenziwa. I-DeepSeek R1, kwa-Perplexity, inikeza ithuluzi elinamandla neliguquguqukayo elivumela abafundi, ochwepheshe, kanye nabafisa ukwazi ukuthola izimpendulo ezinemininingwane, benze ucwaningo olunzulu, noma baxazulule izinselele zobuchwepheshe ngaphandle kwemikhawulo ebekwe amamodeli wangaphambilini.