La Ubuhlakani bokwenziwa kanye nokunyakaza komthombo ovulekile Bashintshe ngokuphelele indlela izinkampani ezakha ngayo isofthiwe kanye nezinhlelo zokusebenza ezenziwe ngokwezifiso. Lokho okwakugcinelwe izinkampani ezinkulu manje sekufinyeleleka kunoma yiliphi ithimba lobuchwepheshe elinokuzimisela okuncane kokushintshashintsha, indawo yokugcina enhle ye-GitHub, kanye nokuhlela okuyisisekelo kokuphepha kwe-inthanethi. Namuhla, kungenzeka ngokuphelele "ukukopisha" (noma, ukuhlanganisa) izinhlelo zokusebenza ezisekelwe ku-AI, uzifake kumaseva akho, futhi uzivumelanise nezinqubo zakho.
Kulesi sihloko sizobuyekeza ngokuningiliziwe amaphrojekthi obuhlakani bokwenziwa alungele ukwenziwa ama-clone futhi usebenzise: kusukela ekubhalweni ngezwi kuya kuma-ejenti azimele, abasizi bendawo, izinjini zokusesha amadokhumenti ezinkampani, ukwenziwa kwezithombe, ukwenziwa kwezwi, noma abalingani bokuhlela izinhlelo. Uzobona nokuthi inkampani yokuthuthukisa efana ne-Q2BSTUDIO ingakusiza kanjani ukuguqula la mathuluzi abe izixazululo zesofthiwe zobungcweti, ezenziwe ngokwezifiso ezihlanganiswe nezinsizakalo zefu ze-AWS ne-Azure, ubuhlakani bebhizinisi, kanye ne-Power BI.
Ukukopisha izinhlelo zokusebenza ezisebenzisa i-AI: kusho ukuthini ngempela amaphrojekthi "okwenza ama-cloning"
Uma sikhuluma ngakho kopisha izinhlelo zokusebenza ngobuhlakani bokwenziwa Asikhulumi ngokuqola isofthiwe noma ukukopisha imikhiqizo ekhokhelwayo, kodwa kunalokho sisebenzisa amaphrojekthi omthombo ovulekile ashicilelwe ku-GitHub noma kwamanye amapulatifomu akuvumela ukuthi ukopishe ikhodi yawo ngokukhululeka. Lawa ma-repository ngokuvamile afaka yonke i-AI logic, imiyalelo yokufaka, futhi, ezimweni eziningi, izibonelo ezenziwe ngomumo ezingashintshwa ukuze zivumelane namabhizinisi ahlukene.
Inzuzo enkulu ukuthi ungakwazi faka la maphrojekthi kumaseva akhoKungakhathaliseki ukuthi kuse-site noma efwini (isb., ku-AWS noma ku-Azure), lawula idatha futhi wenze ngezifiso uhlelo lokusebenza njengengxenye yesofthiwe yakho yangokwezifiso: shintsha isikhombimsebenzisi, xhuma kuzizindalwazi zakho, chaza ukugeleza kwangaphakathi, noma ukuhlanganise namathuluzi akho obuhlakani bebhizinisi namadeshibhodi ku-Power BI.
Ngamazwi asebenzayo, "ukukopisha" uhlelo lokusebenza lwe-AI kuvame ukuhilela hlanganisa indawo yokugcina nge-GitLungisa indawo (i-Python, imitapo yolwazi, amamodeli, izitsha…), landela imiyalelo yokusetshenziswa, bese uqala lapho, uthuthukisa okwengeziwe. Izinkampani ezigxile kwizicelo ezenziwe ngokwezifiso ezifana ne-Q2BSTUDIO Bangakwazi ukusingatha konke lokhu kwezobuchwepheshe kanye nokuhlanganiswa nezinhlelo zakho zebhizinisi, besebenzisa izindlela ezinhle zokuphepha kwe-inthanethi kanye nemikhuba yokukhulisa.
I-OpenAI's Whisper: bhala umsindo ube umbhalo ngokunemba okuphezulu
I-Whisper iyimodeli yokuqaphela izwi Yasungulwa yi-OpenAI, ivelele ngokunemba kwayo kanye namakhono ayo okusebenzisa izilimi eziningi. Ilungele ukubhalwa kwama-podcast, izingxoxo, ama-webinar, izinkulumo, imihlangano yeqembu, noma yikuphi ukuqoshwa komsindo inkampani yakho ekukhiqiza nsuku zonke.
Ukufakwa kwayo okujwayelekile ezindaweni zePython kulula njengokusebenzisa i- ukufaka i-pip openai-whisper (noma okufanayo kuye ngenguqulo kanye nokupakishwa kwamanje). Ngemuva kwalokho, ungayiphakela amafayela alalelwayo bese uthola imibhalo ecacile, elungele ukukhomba, ukuhlaziya, noma ukuyifaka emisebenzini yakho yobuhlakani bebhizinisi.
Ku-GitHub, indawo yokugcina i-Whisper esemthethweni (github.com/openai/whisper) ifaka izibonelo zokusetshenziswa, imibhalo, kanye nemingcele yokucushwa. Ngokuhlanganisa le phrojekthi ngendlela efanayo, ungayihlanganisa nezinhlelo zakho zokusebenza ezenziwe ngokwezifiso.Kusukela kumaphaneli angaphakathi okulayisha umsindo womhlangano nokukhiqiza amaminithi azenzakalelayo, kuya ezinhlelweni eziguqulayo imihlangano ye-intanethi kokuqukethwe okungasetshenziswa kabusha kokumaketha noma ukuqeqeshwa.
Esimweni senkampani, inhlanganisela enamandla kakhulu iwukuhlanganisa i-Whisper izinsizakalo zobuhlakani bebhizinisiAma-transcript agcinwa echibini lakho ledatha noma kusizindalwazi, afakwe ohlwini, bese efinyelelwa kusetshenziswa amathuluzi afana ne-Power BI noma izinjini zokusesha zebhizinisi ezithuthukisiwe. Lokhu kuvumela ithimba lakho ukuthola ngokushesha lokho okushiwo emhlanganweni othize, ukuthi yiziphi izivumelwano ezifinyelelwe, noma ukuthi yiziphi izihloko ezixoxwe ngazo kumakomidi athile.
I-AutoGPT: ama-ejenti azimele okuzenzakalelayo kwemisebenzi eyinkimbinkimbi
I-AutoGPT ingenye yama-ejenti e-AI azimele aziwa kakhuluIsebenzisa amamodeli ohlobo lwe-GPT ukulandelelanisa izenzo, ukuhlela, kanye nokwenza imisebenzi ngaphandle kokugadwa okuqhubekayo, njalo ngaphakathi kwemingcele oyibekile. Esikhundleni sokucela impendulo eyodwa, ubeka umgomo obanzi, bese umenzeli ehlukanisa lowo mgomo ube yizinyathelo ezincane aziqedayo.
Ukuze uqale ukusebenza, ngokuvamile udinga I-Python ifakiwe, ukuncika okuthile, kanye nokhiye we-API ngemodeli yolimi ofuna ukuyisebenzisa (isibonelo, evela ku-OpenAI noma kwabanye abahlinzeki abasekelwayo). Indawo yokugcina eyinhloko itholakala ku-github.com/Significant-Gravitas/Auto-GPT, lapho izidingo zobuchwepheshe, iziguquguquko zemvelo, kanye nezindlela zokwenza zichazwe kabanzi.
Ezinkampanini, i-AutoGPT ifanelana kahle kakhulu ne- imisebenzi yokumaketha kanye nokukhiqizwa kokuqukethwe: lungiselela okusalungiswa kwezihlokoImikhankaso ye-imeyili, iziphakamiso zokuthengisa, noma izifinyezo ezisekelwe olwazini lwangaphakathi. Ingasebenza futhi njenge-ejenti ye-AI eqhuba ucwaningo oluyisisekelo, ibukeze idatha yomphakathi, noma yakhe amadokhumenti okuqala amaphrojekthi esofthiwe enziwe ngokwezifiso.
Kodwa-ke, uma usebenzisa ama-ejenti azimele, kungcono ukusungula imingcele ecacile yokuphepha kanye nokuphepha kwe-inthanethiKhawulela ukufinyelela, chaza ukuthi yiziphi izinhlelo ezingafinyelelwa, lawula ukusetshenziswa kwe-API, futhi uqaphe imiphumela ngaphambi kokuyenza ibe sesidlangalaleni. Umlingani wezobuchwepheshe ofana ne-Q2BSTUDIO angakusiza ukuthi uhlanganise i-AutoGPT ezinqubweni zakho futhi uyixhume ezinhlelweni zakho zamafu ku-AWS kanye ne-Azure ngaphandle kokubeka idatha ebucayi engcupheni.
I-GPT4All: abasizi bendawo ngaphandle kokuthembela efwini
cunt I-GPT4All ikuvumela ukuthi usebenzise amamodeli ohlobo lwe-GPT kukhompyutha yakho. o amasevangisho nangesibonisi esilula sezithombe. Kuyathakazelisa kakhulu izinhlangano ezifuna ukuzama ama-chatbot nabasizi bangaphakathi, kodwa ezingafuni ukuthi idatha yazo yabelwane nezinsizakalo zangaphandle ngezizathu zobumfihlo noma zokuthobela imithetho.
Le phrojekthi igxile ekuhlinzekeni amamodeli alungiselelwe kahle angasebenza kuhadiwe encane, enezinguqulo ezahlukene kuye ngokuthi uhlose ukuyisebenzisa kanjani (umsizi wenhloso ejwayelekile, umsizi ogxile kukhodi, njll.). Ngokuhlanganisa indawo yokugcina ye-GPT4All (github.com/nomic-ai/gpt4all), ungalanda amamodeli, ulungiselele indawo, bese uqalisa isikhombimsebenzisi.
Kwamabhizinisi, lokhu kuvula ithuba abasizi be-AI bangaphakathi abangaxhomekile ku-inthanethiIsibonelo, i-chatbot esiza abasebenzi ngezinqubo zenkampani, imibuzo yangaphakathi evame ukubuzwa, izinqubomgomo zokuphepha kwe-inthanethi, noma imibuzo mayelana namathuluzi ezinkampani. Konke lokhu kusebenza kumaseva akho, ngemithetho yakho yokusekela, ukuqapha, kanye nokulawula ukufinyelela.
Ngaphezu kwalokho, i-GPT4All ihlangana kahle kakhulu ne- izixazululo zesofthiwe ezenziwe ngokwezifisoInkampani efana ne-Q2BSTUDIO ingakha amadeshibhodi ayo lapho imodeli iphendula khona ngokusekelwe ezisekelweni zakho zolwazi, iyixhume kumasevisi akho efu (i-AWS, i-Azure) ukuze igcine izingxoxo, futhi iyixhume ezinjinini zokusesha ezihlanganisa i-AI nedatha yakho yebhizinisi ehlelekile.
I-PrivateGPT: Buza imibuzo ngamadokhumenti akho ngaphandle kokushiya indawo okuyo
I-PrivateGPT yakhelwe ukuphendula imibuzo ngokusekelwe kumadokhumenti endawo (njenge-PDF, izinkontileka, imibiko, izincwadi zangaphakathi, noma ama-imeyili athunyelwe ngaphandle) ngaphandle kokuthumela ulwazi efwini. Lena indlela ethakazelisa kakhulu yezomthetho, ukuthobela imithetho, iminyango yezabasebenzi, noma kunoma iyiphi indawo ephatha idatha ebucayi.
Indlela evamile ilula: ukopisha indawo yokugcina (github.com/imartinez/privateGPT), ufake ama-dependence, ufake amadokhumenti akho kumafolda akhonjisiwe, bese uhlelo lukhiqiza ama-index adingekayo ukuze kwenziwe imibuzo ngolimi lwemvelo. Konke kusebenza endaweni yakini, okusiza ukuthobela izinqubomgomo zokuphepha kwe-inthanethi zangaphakathi.
Nge-PrivateGPT, ithimba lezomthetho lingalayisha izinkontileka futhi libuze imibuzo ethile (isibonelo, izinsuku zokugcina zokuvuselela, izigaba zobumfihlo, noma izinhlawulo). Ithimba lokusebenza lingalayisha amabhukwana emishini futhi libuze ukuthi lingayixazulula kanjani inkinga ethile. Isihluthulelo ukuthi Ulwazi luhlala ezinhlelweni zakho, ngaphandle kokuncika kuma-API angaphandle.
Ukuhlanganisa i-PrivateGPT kwisisombululo senkampani esenziwe ngokwezifiso kuvumela, isibonelo, ukwengeza ukuqinisekiswa okusekelwe ezindimeni, ukuhlolwa kwemibuzo, ukuhlanganiswa nezindawo zokugcina amadokhumenti, kanye namadeshibhodi e-Power BI ahlaziya ukuthi yiziphi izihloko ezivame ukukhulunywa ngazo kanye nokuthi kukhona izikhala zamadokhumenti.
I-Stable Diffusion WebUI AUTOMATIC1111: Izithombe ezikhiqizwe yi-AI zebhizinisi lakho
Ukuhlanganiswa kwe Ukusabalalisa Okuzinzile nge-AUTOMATIC1111 WebUI Sekuyindinganiso yangempela yokukhiqiza izithombe ezivela ezincazelweni zombhalo. Lesi sikhombikubona sezithombe senza ukusebenzisa imodeli kube lula kakhulu: ukhetha isifinyezo, izilungiselelo eziyisisekelo, imodeli, isixazululo, futhi ngemizuzwana nje uneziphakamiso ezibonakalayo.
Enye yamandla alo msebenzi ukuthi, ezimweni eziningi, ingasetshenziswa ne Abafaki bohlobo "lokuchofoza kanye" kumishini ehambisanayo, okusheshisa ukuhlolwa kanye nokufakwa kokuqala. Indawo yokugcina esemthethweni itholakala ku-github.com/AUTOMATIC1111/stable-diffusion-webui futhi ifaka imiyalelo yezinhlelo zokusebenza ezahlukene.
Ezingeni lebhizinisi, leli thuluzi lilungele kahle Dala imidwebo yomkhiqizo, imiqondo yomkhiqizo, amabhanela, kanye nezinsiza zokumaketha ngesikhathi esifushane. Amaqembu okuklama angakha imibono eminingi, athuthukise leyo ayithola ithakazelisa kakhulu, bese esebenza kuyo ngamathuluzi awo avamile.
Ihlanganiswe nesofthiwe eyenziwe ngokwezifiso, i-WebUI noma izingxenye zayo zingaba yingxenye yamaphothali amakhasimende (isibonelo, ukuze ziwavumele ukuthi abuke iziphakamiso ezenziwe ngokwezifiso) noma izinhlelo zokukhiqiza okuqukethwe kwangaphakathi. Konke lokhu kuxhunywe engqalasizinda yakho yamafu (i-AWS, i-Azure) futhi kunamathela ezinqubomgomo zokuphepha kwe-inthanethi kanye nokuphathwa kwempahla yedijithali yenhlangano yakho.
I-Deepset Haystack: Izinjini zokusesha ezihlakaniphile kudatha yakho
I-Haystack ilayibrari eyenzelwe ukwakha izinjini zokusesha kanye nezinhlelo zemibuzo nezimpendulo ezisebenza kumadokhumenti angaphakathi noma emithonjeni yedatha. Ivumela ukuhlanganisa ama-backend ahlukene (i-ElasticSearch, i-OpenSearch, njll.), amamodeli olimi, kanye nokucubungula amapayipi ukuze kunikezwe izimpendulo ezinembile nezihambisana nomongo.
Ngokuhlanganisa indawo yokugcina i-Haystack (github.com/deepset-ai/haystack) ungasetha noma yini kusukela kusiphequluli sebhizinisi esidala kuya ku- Umsizi wohlobo "lwemibuzo nezimpendulo" mayelana nedokhumenti yakhoIwusizo ikakhulukazi ezinkampanini ezinenqwaba yolwazi olusabalele kuma-wiki, izinhlelo zokuphatha amadokhumenti, amathuluzi okuthengisa amathikithi, kanye nezindawo zokugcina amafu.
Ezindaweni zobuhlakani bebhizinisi, i-Haystack ingasebenza njengesendlalelo sokufinyelela olwazini olungahlelekile, ihambisana namadeshibhodi e-Power BI noma izixazululo ezifanayo. Abasebenzisi babuza imibuzo ngolimi lwemvelo. (“Yiziphi izinguquko ezazikhona kunqubomgomo yamaholide ngo-2023?”) futhi uhlelo luthola amadokhumenti afanele, lufingqa okuqukethwe.
Ukuhlanganisa i-Haystack kuzinhlelo zokusebenza ezenziwe ngokwezifiso kukuvumela ukuthi udale okuhlangenwe nakho kokusesha okuthuthukile kakhulu, ngezihlungi ngendima, umnyango, ulimi, noma uhlobo lokuqukethwe. I-Q2BSTUDIO, isibonelo, ingaxhuma i-Haystack ezinhlelweni zakho ku-AWS ne-Azure, kanye nokugeleza kwedatha yakho ebalulekile, ukuqinisekisa ukuqinisekiswa, ukubhaliswa, kanye nokuhambisana nezidingo zakho zokuphepha kwe-inthanethi.
Ukwenziwa Kwezwi Ngesikhathi Sangempela: Ukwenziwa Kwezwi Okunomthwalo Wemfanelo
I-Real Time Voice Cloning iphrojekthi ye-Python Ikuvumela ukuthi ukhiqize amazwi okwenziwa ngemizuzwana embalwa nje yomsindo obhekiselwe kuwo. Ihlaba umxhwele ngobuchwepheshe, futhi inozwela kakhulu ngokombono wokuziphatha nowomthetho, ngakho-ke ukusetshenziswa kwayo okunesibopho kubalulekile.
Indawo yokugcina impahla eyinhloko (github.com/CorentinJ/Real-Time-Voice-Cloning) ikhombisa indlela yokuqeqesha nokusebenzisa amamodeli fanisa ama-timbre ezwi futhi udale ama-voiceover asekelwe kumbhalo. Lokhu kukuvumela ukuthi udale abasizi bezwi abanethoni ethile, imiyalezo ezenzakalelayo, noma imiboniso yezobuchwepheshe yezindawo zesevisi yamakhasimende.
Ezimweni zenkampani, ukusetshenziswa okunconywayo kufaka phakathi amazwi abonakala ngokucacile njengawokwenziwa noma ngemvume ecacile. Isibonelo, Imisindo yezinhlelo ze-IVR, izimemezelo zangaphakathi, noma abasizi ababonakalayo ezikhonza abasebenzi namakhasimende. Zingasetshenziswa futhi kumaphrojekthi okufinyeleleka, ukuqeqeshwa, noma ukwenza izibonelo zomkhiqizo.
Kulezi zinhlobo zezixazululo, ukuphepha kwe-inthanethi kanye nezinqubomgomo zokusebenzisa kubalulekile: ukusungula ukuthi ubani ongaqeqesha amamodeli, ngedatha yezwi, nokuthi ngaziphi izinjongo. Izinkampani ezigxile ku-AI yebhizinisi zingakusiza ukuklama izinhlaka zokuphatha, izivikelo zobuchwepheshe, kanye nezilawuli zokufinyelela ukuvimbela ukusetshenziswa kabi.
I-OpenDevin: I-AI esebenzela ukuthuthukiswa kwesofthiwe ngokwezifiso
I-OpenDevin isebenza njengomsizi wokuhlela Ikhiqiza ikhodi, izikripthi, kanye nezixazululo zobuchwepheshe ezivela emiyalweni yolimi lwemvelo. Luhlobo "lomthuthukisi obonakalayo" olusiza ithimba lakho labantu ukuthi lisebenze ngokushesha emisebenzini ephindaphindwayo noma inguqulo yokuqala yezici ezithile.
Ngemva kokwenza i-repository ibe yi-clone (github.com/OpenDevin/OpenDevin) futhi ulungiselele ama-API noma amamodeli adingekayo, ungacela ithuluzi ukuthi libhale iziqeshana zekhodi, lenze izivivinyo zenzeka ngokuzenzakalela, lidale izikripthi zokufaka, noma liphakamise izakhiwo zephrojekthi. Akuthathi indawo yabathuthukisikodwa kuyabakhulula kweminye yemisebenzi yemishini.
Kumaphrojekthi esofthiwe enziwe ngokwezifiso, i-OpenDevin ikuvumela ukuthi unciphise isikhathi sokuthuthukiswa kwamamojula ajwayelekile (ukugunyazwa, amaphaneli e-CRUD, ukuhlanganiswa nama-API ajwayelekile, njll.), okusiza ithimba lakho ukuthi ligxile ku-logic yebhizinisi ehlukanisayo. Lokhu kuhunyushwa kube ukulethwa okusheshayo kanye nokuphindaphinda okwengeziwe nekhasimende lokugcina.
Ihlanganiswe namapayipi e-CI/CD, izindawo zokugcina amafu (isb., i-AWS CodeCommit, i-Azure DevOps, noma i-GitHub Enterprise), kanye namathuluzi okuphatha amaphrojekthi, i-OpenDevin ingaba yingxenye yesistimu yobunjiniyela yenkampani yakho, njalo ngaphansi kokuqondisa konjiniyela abangochwepheshe abaqinisekisa ikhodi ekhiqiziwe.
ULeon: Umsizi wezwi lomuntu siqu osebenza endaweni yakini
ULeon ungumsizi womuntu siqu womthombo ovulekile Ilawulwa yizwi futhi yakhelwe ukusebenza kumadivayisi akho, ngaphandle kokuthembela kumapulatifomu angaphandle. I-modular, ngakho ungayikhulisa ngamaphakheji enziwe ngokwezifiso anezela amakhono amasha nezixhumi.
Ikhodi kaLeon iyatholakala ku-github.com/leon-ai/leon futhi ivumela ukuthi ifakwe njengengxenye eyinhloko ye izixazululo zokukhiqiza kwabasebenzi: izikhumbuzi, ukuvula izinhlelo zokusebenza zangaphakathi, ukubonisana ngolwazi oluyisisekelo, ukuhlanganiswa namakhalenda noma ngisho nokwenza imisebenzi ethile yokusebenza.
Njengoba esendaweni futhi enwebeka kalula, uLeon ufanelana kahle kakhulu nezimo lapho kudingeka khona ukusebenzisa izwi njengesixhumi esibonakalayo ukusebenzisana nezinhlelo zokusebenza zebhizinisi, kodwa ngaphandle kokudalula idatha kubasizi bebhizinisi benkampani yangaphandle. Ngokuhlanganisa kahle nezinsizakalo zefu ze-AWS ne-Azure, ungavumelanisa idatha, uqalise imisebenzi, futhi uxhume kuma-API angaphakathi.
Ngaphakathi kwephrojekthi yesicelo esenziwe ngokwezifiso, inkampani efana ne-Q2BSTUDIO ingakha amamojula athile kaLeon afanelana nomkhakha wakho: kusukela kubasizi babaqhubi bezitshalo kuya ekusekeleni ithimba lokuthengisa, njalo enakekela ukuqinisekiswa, ukuhlolwa kanye nokulandelela ukusetshenziswa ukuze ahambisane nezinqubomgomo zakho zokuphepha.
i-llama.cpp: Amamodeli e-LLaMA alungiselelwe i-CPU
cunt I-llama.cpp ikuvumela ukuthi usebenzise amamodeli avela emndenini we-LLaMA (nezinye izinhlelo ezihambisanayo) ngempumelelo ku-CPU, ngisho nakumadivayisi angenayo i-GPU enamandla. Le phrojekthi ihlose ukunikeza ukusetshenziswa okulungiselelwe kakhulu, nge-quantization kanye namasu afanayo, ukuze i-AI yengxoxo itholakale kumadivayisi aphansi.
Indawo yokugcina eyinhloko (github.com/ggerganov/llama.cpp) ifaka amathuluzi okuguqula amamodeli, ukuwasebenzisa kusuka kulayini womyalo, noma ukuwahlanganisa kuzinhlelo zokusebenza ezinezixhumi ezihlukile. Lokhu kwenza kube lula ukusetha ama-chatbot endawo., abasizi bokusekela noma ama-ejenti angaxhunyiwe ku-inthanethi asebenza ngaphandle kokuxhumeka kwefu.
Kumabhizinisi aqaphela ubumfihlo, i-llama.cpp iyisisekelo esihle kakhulu sokwakha phezu kwaso. ama-prototype e-AI engxoxo kanye nezixazululo okumele zisetshenziswe ezindaweni ezikude noma ezinokuxhumana okulinganiselwe (isb., emafektri, ezikhungweni ezikude, noma ezindaweni ezinezidingo eziqinile zokuphepha kwe-inthanethi).
Ngokuhlanganisa i-llama.cpp kusofthiwe eyenziwe ngokwezifiso, ungakha izixhumi zewebhu zangaphakathi, izinhlelo zokusebenza zedeskithophu, noma izinsizakalo ezisingatha ngokushesha imibuzo yangaphakathi. Uma ihlanganiswe ne-PrivateGPT noma i-Haystack, iba injini yolimi echaza imibuzo ngenkathi ezinye izingxenye ziphatha amadokhumenti nedatha yebhizinisi.
Ukuhlaziywa kwedokhumenti okusebenzisa i-Base44 kanye ne-AI ukudala izinhlelo zokusebenza ezingenayo ikhodi
Ngaphezu kwamaphrojekthi aqoshwe ngqo kusuka ku-GitHub, kunezinkundla ezifana nalezi I-Base44, eyenzelwe ukudala izinhlelo zokusebenza zebhizinisi ngendlela elulaLeli thuluzi lisebenziseka ngezindlela eziningi ngokwanele ukuthuthukisa izinhlelo zokusebenza zokukhiqiza zomuntu siqu, izinsiza zasehhovisi elingemuva, amaphothali amakhasimende, noma izixazululo zangaphakathi zokwenza ngokuzenzakalelayo.
Ifilosofi ye-Base44 ihambisana kahle nomqondo wokuthi Yakha ama-MVP kanye nama-prototypes asheshayo: qala inguqulo yokuqala esebenzayo yethuluzi, uqinisekise ngabasebenzisi bangempela, bese, kusukela lapho, unquma ukuthi yiziphi izingxenye okufanele ziguquke zibe isofthiwe eqinile noma zihlanganiswe nezinye izingcezu ze-AI zomthombo ovulekile.
Iphuzu elilodwa elithakazelisa kakhulu yilo umhlaziyi wedokhumenti onobuhlakani bokwenziwaLeli thuluzi likuvumela ukuthi uguqule ama-PDF, ama-imeyili, kanye namadokhumenti askeniwe abe idatha ehlelekile. Ngamanye amazwi, lapho kwakukhona khona ifayela "elifile" elalinzima ukulisebenzisa, manje sekunamasimu achazwe kahle alungele ukondla izinhlelo zakho.
Le datha ehlelekile ingathunyelwa kwezinye izinhlelo zokusebenza ngaphandle kwesidingo sokuhlelaLokhu kungafezwa ngokuhlanganiswa okubonakalayo noma izixhumi ezichazwe kusengaphambili, noma ngokuzifaka emikhiqizweni yakho ngama-API. Ngale ndlela, ungenza ukufakwa kwedatha kube ngokuzenzakalela, unciphise amaphutha abantu, futhi usheshise izinqubo ezazidinga ukungenelela okukhulu ngesandla ngaphambilini.
Ngokuhlanganisa amathuluzi afana ne-Base44 namaphrojekthi omthombo ovulekile okukhulunywe ngawo ngenhla, izinkampani zifinyelela ibhalansi ethakazelisa kakhulu: isivinini sokubhala kanye nokuqina kokulinganisaI-Q2BSTUDIO ingahlela le nhlanganisela, iklame izakhiwo ezisebenzisa amapulatifomu anekhodi ephansi noma angenayo ikhodi lapho kunengqondo khona kanye nekhodi yangokwezifiso lapho ukwenza ngokwezifiso noma ukusebenza kudinga khona.
Indlela yokukhetha iphrojekthi ye-AI efanele yenkampani yakho
Njengoba kunezinketho eziningi kangaka etafuleni, kunengqondo ukuzibuza Yimuphi umsebenzi we-AI ofanelana kahle nemigomo yakhoUkukhetha kuncike, ngaphezu kwakho konke, enkingeni ofuna ukuyixazulula kanye nemingcele yendawo yakho yezobuchwepheshe neyokulawula.
Uma okudingayo guqula umsindo ube umbhaloKungakhathaliseki ukuthi kungokwemibhalo yangaphakathi, ukuqeqeshwa, noma ukuhlaziywa kwangemuva, i-Whisper iyona engcono kakhulu. Ngemisebenzi lapho ufuna i-AI ihlele futhi yenze khona izenzo eziningi zilandelana (njengemikhankaso yokuqukethwe noma imisebenzi yocwaningo oluyisisekelo), i-AutoGPT iyindlela ecacile.
Uma kuqala kwakho Xoxa nomodeli endaweni yangakini ngaphandle kokuthembela efwiniI-GPT4All noma i-llama.cpp zingabangane abahle kakhulu: eyokuqala igxile kubasebenzisi bokugcina abane-interface yesithombe, kanti eyesibili igxile ekusebenzeni kwe-CPU. Ukuze ubuze ngemibhalo yakho yangaphakathi ngenkathi ugcina amafayela ngaphansi kolawulo lwakho, i-PrivateGPT noma i-Haystack (noma kokubili) yizinketho okufanele uzikhethe.
Embonini ebonakalayo, i-Stable Diffusion WebUI AUTOMATIC1111 ihlangabezana nezidingo ze- Ukwenziwa kwezithombe zokumaketha, ukwenza ama-prototyping, noma ukucabangaEzimweni lapho izwi libalulekile, i-Real Time Voice Cloning ivumela ukuzama ama-clone ezwi inqobo nje uma izimiso zokuziphatha nezomthetho zihlonishwa.
Ukuze kusheshiswe umnyango wakho wokuthuthukisa, i-OpenDevin iba umsizi womshayeli wendiza osiza ngokubhala ikhodi kanye nokwenza izinto ngokuzenzakalela, kuyilapho uLeon esebenza njengomshayeli wendiza. umsizi wezwi lomuntu siqu wokukhiqiza kwangaphakathiUma ufuna futhi ukwakha izinhlelo zokusebenza zebhizinisi ngokushesha ngokugeleza kwedatha okuhlelekile, i-Base44 kanye nesihlaziyi sayo sedokhumenti esisebenza nge-AI sinikeza isisekelo esiqinile kakhulu.
Indlela i-Q2BSTUDIO engakusiza ngayo ukuletha lawa ma-AI ekukhiqizweni
Konke okungenhla kuzwakala kukuhle, kodwa iqiniso liwukuthi ukusuka endaweni yokugcina i-GitHub kuya kwisisombululo sebhizinisi lokukhiqiza Akuyona into encane. Yilapho ulwazi lwezinkampani zokuthuthukisa ezikhethekile ezifana ne-Q2BSTUDIO luqala khona ukusebenza, luhlanganisa ulwazi kuzinhlelo zokusebenza ezenziwe ngokwezifiso, isofthiwe eyenziwe ngokwezifiso, ubuhlakani bokwenziwa, kanye nokuphepha kwe-inthanethi.
I-Q2BSTUDIO ingahlaziya izidingo zakho futhi iphakamise ukwakheka okuhlanganisa la maphrojekthi omthombo ovulekile ngezinhlelo zakho ezikhona, kungakhathaliseki ukuthi zigcinwe ezikhungweni zakho zedatha noma kumasevisi efu le-AWS ne-Azure. Baphatha ukufakwa, ukuhlelwa, ukuthunyelwa kweziqukathi uma kudingeka, kanye nokuqapha ezindaweni zangempela.
Ngaphezu kwalokho, banolwazi ku- izinsizakalo zobuhlakani bebhizinisiLokhu kusho ukuthi bangathatha imiphumela evela kulezi zinhlelo ze-AI (ama-Whisper transcripts, izimpendulo ze-PrivateGPT, idatha ehlelekile ye-Base44, njll.) bese beyifaka kumamodeli wedatha yokuhlaziya angasetshenziswa ngaphakathi kwe-Power BI. Lokhu akugcini nje ngokwenza imisebenzi ngokuzenzakalela kodwa futhi kunikeza ukuqonda okusebenzayo.
Ngokuphathelene nokuphepha, i-Q2BSTUDIO ingakusiza ukuthi usebenzise izilawuli zokufinyelela, ukubethela, ukuhlukaniswa kwenethiwekhi, ukuhlolwa, kanye nokuhambisana nemithethoLokhu kubaluleke kakhulu ezixazululweni ze-AI ezisingatha ulwazi olubucayi, njengezindawo zokugcina amadokhumenti angaphakathi, izinkontileka, idatha yamakhasimende, noma ukuqoshwa kwezwi.
Okokugcina, indima yayo njengomhlanganisi ivumela zonke lezi zingcezu ukuthi zisebenze njenge uhlelo lwe-AI oluhambisanayo lwamabhizinisiAma-ejenti e-AI, abasizi bezwi, abakhiqizi bezithombe, izinjini zokusesha ezihlakaniphile, namathuluzi axhumene anekhodi ephansi, ahambisana nezinqubo zakho kanye nesiko lobuchwepheshe lenhlangano yakho.
Ukuhlanganiswa kwamaphrojekthi e-AI omthombo ovulekile, amapulatifomu afana ne-Base44, kanye nokusekelwa komlingani wezobuchwepheshe ofana ne-Q2BSTUDIO kuvula isimo lapho ukukopisha, ukulungisa, kanye nokuhlanganisa izinhlelo zokusebenza zobuhlakani bokwenziwa kuphelelwa yisikhathi kube yindlela yokuzihlola ehlukanisiwe futhi kuba isu langempela lokuthola ukusebenza kahle, ukusungula izinto ezintsha ngokushesha, nokwenza izinqumo zebhizinisi ezihlakaniphile ezisekelwa yizinsizakalo zamafu ze-AWS kanye ne-Azure, izinsizakalo zobuhlakani bebhizinisi, kanye nezixazululo ezithuthukisiwe nge-Power BI.
