{"id":1997,"date":"2024-05-14T10:54:18","date_gmt":"2024-05-14T10:54:18","guid":{"rendered":"https:\/\/howtogeek.blog\/lv\/?p=1997"},"modified":"2024-05-14T10:54:18","modified_gmt":"2024-05-14T10:54:18","slug":"whats-new-in-chatgpt-with-gpt-4o-release-lv","status":"publish","type":"post","link":"https:\/\/howtogeek.blog\/lv\/whats-new-in-chatgpt-with-gpt-4o-release-lv\/","title":{"rendered":"Kas jauns pakalpojum\u0101 ChatGPT ar GPT-4o laidienu?"},"content":{"rendered":"

OpenAI nesen pazi\u0146oja par savu GPT-4o modeli\u00a0\u2014 jaunu ChatGPT dzin\u0113ja vado\u0161o modeli\u00a0\u2014 t\u0101 slaveno AI darbin\u0101tu t\u0113rz\u0113\u0161anas robotu, kas pirmo reizi tika izlaists 2022.\u00a0gad\u0101. \u0160is jaun\u0101kais modelis cilv\u0113ka un datora mijiedarb\u012bbu pace\u013c piln\u012bgi jaun\u0101 l\u012bmen\u012b, padarot to \u0101tr\u0101ku un \u0101tr\u0101ku. daudz dabisk\u0101ka.<\/p>\n

GPT-4o ‘o’ apz\u012bm\u0113 ‘omni’, jo tas apvieno tekstu, att\u0113lu un audio vien\u0101 model\u012b. Lai gan inteli\u0123ences un argument\u0101cijas front\u0113 nav uzlabojumu sal\u012bdzin\u0101jum\u0101 ar GPT-4 Turbo modeli, ir daudz jaunu atjaunin\u0101jumu. Tas ir izstr\u0101d\u0101ts, lai nodro\u0161in\u0101tu \u0101tr\u0101kas, cilv\u0113cisk\u0101kas atbildes un pat var simul\u0113t emocijas. Tas ar\u012b iev\u0113rojami \u0101tr\u0101k saprot vizu\u0101l\u0101s un audio ievades. \u0160aj\u0101 dzi\u013caj\u0101 nir\u0161anas laik\u0101 m\u0113s apskat\u012bsim GPT-4o pied\u0101v\u0101t\u0101s funkcijas un to, k\u0101 tas main\u012bs m\u016bsu mijiedarb\u012bbu ar AI pal\u012bgiem. T\u0101tad, pievienojieties mums un s\u0101ksim!<\/p>\n

GPT-4o ir iev\u0113rojami \u0101tr\u0101ks<\/h2>\n

GPT-4o modelis \u013cauj ChatGPT pie\u0146emt ievadi da\u017e\u0101d\u0101s form\u0101s, tostarp audio, tekstu, att\u0113lus vai to kombin\u0101cijas. Tas var ar\u012b sniegt atbildes da\u017e\u0101dos form\u0101tos. Tom\u0113r visiespaid\u012bg\u0101kais ir \u0101trums, k\u0101d\u0101 tas sniedz jums atbildes.<\/p>\n

Izmantojot jauno modeli, ChatGPT prasa tikai aptuveni 320 milisekundes vai pat maz\u0101k, lai rea\u0123\u0113tu uz audio ievadi, kas ir tuvu laikam, kas nepiecie\u0161ams, lai persona atbild\u0113tu sarun\u0101. Turkl\u0101t GPT-4o veiktsp\u0113ja ir l\u012bdz\u012bga GPT-4 Turbo, cikt\u0101l tas attiecas uz ang\u013cu valodas lieto\u0161anu vai koda rakst\u012b\u0161anu.<\/p>\n

Tas ar\u012b darbojas daudz lab\u0101k, rea\u0123\u0113jot uz teksta ievadi valod\u0101s, kas nav ang\u013cu valod\u0101, un ir iev\u0113rojami l\u0113t\u0101kas API, jo tas ir daudz efekt\u012bv\u0101ks nek\u0101 t\u0101 priek\u0161g\u0101j\u0113js.<\/p>\n

Tam ir ar\u012b jauns interfeiss, kas paredz\u0113ts, lai padar\u012btu mijiedarb\u012bbu viegl\u0101ku un gl\u012bt\u0101ku.<\/p>\n

\"\"
Jauns ChatGPT-4o interfeiss.<\/span><\/figcaption><\/figure>\n

Tas ir multimod\u0101ls<\/h2>\n

ChatGPT eso\u0161ais balss re\u017e\u012bms sast\u0101v no trim mode\u013ciem. Pirmaj\u0101 tiek izmantota teksta ievade un tiek nodro\u0161in\u0101ta izvade t\u0101d\u0101 pa\u0161\u0101 form\u0101t\u0101, bet otr\u0101 audio ievadi p\u0101rv\u0113r\u0161 tekst\u0101, ko izmanto ChatGPT, lai nodro\u0161in\u0101tu teksta izvadi. P\u0113d\u0113jais modelis p\u0101rv\u0113r\u0161 tekstu audio form\u0101t\u0101.<\/p>\n

Tom\u0113r \u0161is dizains rada inform\u0101cijas zudumu, jo modelis nevar tie\u0161i mijiedarboties ar audio ieeju vai par\u0101d\u012bt izvadi, kur\u0101 ir emocion\u0101li elementi. Izmantojot GPT-4o, OpenAI ir izveidojis vienu multimod\u0101lu modeli, kas var saprast da\u017e\u0101du veidu ievades un tie\u0161i apstr\u0101d\u0101t nepiecie\u0161amo izvadi.<\/p>\n

Pa\u0161laik tas atbalsta 50 valodas un b\u016bs pieejams ar\u012b k\u0101 API.<\/p>\n

Tas var simul\u0113t cilv\u0113ka emocijas<\/h2>\n

Kad ChatGPT pirmo reizi tika pazi\u0146ots, viens no liel\u0101kajiem AI t\u0113rz\u0113\u0161anas robota tr\u016bkumiem daudziem cilv\u0113kiem bija tas, ka tas nevar\u0113ja sniegt emocion\u0101las atbildes. Izmantojot GPT-4o, ChatGPT var simul\u0113t cilv\u0113ka emocijas un nodro\u0161in\u0101t piem\u0113rotas atbildes, kas integr\u0113 \u0161\u0101das emocijas.<\/p>\n

OpenAI kop\u012bgotaj\u0101s demonstr\u0101cij\u0101s t\u0113rz\u0113\u0161anas robots var redz\u0113t smejamies, dziedam un pat izr\u0101da sarkasmu, kad tas tiek l\u016bgts. Tas ir daudz augst\u0101ks par to, ko pa\u0161laik sp\u0113j jebkur\u0161 cits AI t\u0113rz\u0113\u0161anas robots, un padar\u012bs lietot\u0101ju mijiedarb\u012bbu daudz pat\u012bkam\u0101ku. Saska\u0146\u0101 ar OpenAI teikto, jaunais modelis \u013caus ChatGPT noteikt ar\u012b lietot\u0101ja noska\u0146ojumu un nodro\u0161in\u0101t adekv\u0101tas atbildes, pamatojoties uz \u0161o inform\u0101ciju.<\/p>\n

Apvienojum\u0101 ar sp\u0113ju izmantot vizu\u0101lo ievadi, sp\u0113ja nodro\u0161in\u0101t emocion\u0101las reakcijas ir viena no iespaid\u012bg\u0101kaj\u0101m funkcij\u0101m, ko pied\u0101v\u0101 jaunais modelis.<\/p>\n

Jaunais modelis var uzglab\u0101t liel\u0101ku inform\u0101cijas daudzumu<\/h2>\n

Pateicoties GPT-4o, ChatGPT tagad var lab\u0101k izprast att\u0113lus, piem\u0113ram, diagrammas un fotoatt\u0113lus, k\u0101 ar\u012b saglab\u0101t vair\u0101k inform\u0101cijas par lietot\u0101ju. Tas noz\u012bm\u0113 ar\u012b atbildes, kas var satur\u0113t liel\u0101kus teksta gabalus, kas var b\u016bt noder\u012bgi vair\u0101kos lieto\u0161anas gad\u012bjumos.<\/p>\n

Piem\u0113ram, varat l\u016bgt t\u0113rz\u0113\u0161anas robotam tagad iztulkot liel\u0101kus teksta fragmentus, un tas var jums pal\u012bdz\u0113t veikt tie\u0161o tulko\u0161anu. T\u0101 k\u0101 taj\u0101 var izmantot vizu\u0101l\u0101s un audio ievades, k\u0101 ar\u012b teksta ievades, varat to izmantot, lai apkopotu inform\u0101ciju no ekr\u0101nuz\u0146\u0113mumiem, fotoatt\u0113liem un citiem att\u0113liem.<\/p>\n

T\u0101 k\u0101 GPT-4o var glab\u0101t liel\u0101ku inform\u0101cijas daudzumu bez nepiecie\u0161am\u012bbas atk\u0101rtot, tas \u013cauj lietot\u0101jiem piedal\u012bties t\u0101l\u0101k\u0101s sarun\u0101s ar AI t\u0113rz\u0113\u0161anas robotu. Izmantojot pla\u0161\u0101ku inform\u0101ciju, lietot\u0101ji var sagaid\u012bt, ka sarunas b\u016bs sare\u017e\u0123\u012bt\u0101kas.<\/p>\n

Aizsardz\u012bbas pas\u0101kumi jaunaj\u0101 model\u012b<\/h2>\n

T\u0101pat k\u0101 eso\u0161ie GPT mode\u013ci, ar\u012b GPT-4o ir izstr\u0101d\u0101ts, \u0146emot v\u0113r\u0101 noteiktus dro\u0161\u012bbas pas\u0101kumus, lai nodro\u0161in\u0101tu dro\u0161\u012bbu. Apm\u0101c\u012bbas dati ir filtr\u0113ti, un mode\u013ca darb\u012bba ir preciz\u0113ta p\u0113c apm\u0101c\u012bbas. OpenAI ir ar\u012b nov\u0113rt\u0113jis AI modeli p\u0113c vair\u0101kiem parametriem, piem\u0113ram, kiberdro\u0161\u012bbas un p\u0101rliecin\u0101\u0161anas, lai izvair\u012btos no nev\u0113lamiem incidentiem.<\/p>\n

Turkl\u0101t izstr\u0101d\u0101t\u0101ji izmantoja vair\u0101k nek\u0101 70 ekspertu pal\u012bdz\u012bbu da\u017e\u0101d\u0101s jom\u0101s, piem\u0113ram, dezinform\u0101cij\u0101 un soci\u0101laj\u0101 psiholo\u0123ij\u0101, lai identific\u0113tu un mazin\u0101tu riskus, kas var\u0113tu rasties vai palielin\u0101ties ar jauno modeli. Vi\u0146i ar\u012b turpin\u0101s uzraudz\u012bt t\u0113rz\u0113\u0161anas robota izmanto\u0161anu, lai nov\u0113rstu n\u0101kotnes riskus, un p\u0113c vajadz\u012bbas veiks nepiecie\u0161am\u0101s darb\u012bbas.<\/p>\n

Ies\u0101c\u0113jiem OpenAI dar\u012bs pieejamu tikai ierobe\u017eotu balss mode\u013cu s\u0101kotn\u0113jo iestat\u012bjumu izv\u0113li. Tas \u0146ems v\u0113r\u0101 ar\u012b lietot\u0101ju sniegt\u0101s atsauksmes, lai uzlabotu modeli un nodro\u0161in\u0101tu piln\u012bgu dro\u0161\u012bbu.<\/p>\n

GPT-4o pieejam\u012bba<\/h2>\n

L\u012bdz \u0161im bija pieejamas divas ChatGPT versijas\u00a0\u2014 bezmaksas versija, kas darbojas ar GPT\u00a03.5, un maksas versija, kas darboj\u0101s ar GPT\u00a04.0 un maks\u0101ja 20\u00a0$ m\u0113nes\u012b. P\u0113d\u0113jais var piek\u013c\u016bt liel\u0101kam valodas modelim, kas \u013cauj apstr\u0101d\u0101t liel\u0101ku datu apjomu.<\/p>\n

GPT-4o b\u016bs pieejams gan bezmaksas, gan maksas lietot\u0101jiem, un paredz\u0113ts, ka tas tiks ieviests ier\u012bc\u0113s tuv\u0101ko ned\u0113\u013cu laik\u0101. T\u0101 b\u016bs pieejama mobilaj\u0101s ier\u012bc\u0113s, izmantojot eso\u0161o lietotni, kas tiks atjaunin\u0101ta, un papildus t\u012bmek\u013ca versijai galddatoru lietot\u0101jiem b\u016bs pieejama ar\u012b jauna macOS darbvirsmas lietotne (Windows versija b\u016bs pieejama v\u0113l\u0101k \u0161ogad).<\/p>\n

Lai at\u0161\u0137irtu bezmaksas un maksas lietot\u0101jus, OpenAI maksas lietot\u0101jiem pied\u0101v\u0101s piecas reizes liel\u0101ku jaudu, nek\u0101 tiks pied\u0101v\u0101ts bezmaksas lietot\u0101jiem. Kad b\u016bsiet izt\u0113r\u0113jis limitu, tiksiet p\u0101rsl\u0113gts atpaka\u013c uz ChatGPT\u00a03.5.<\/p>\n

OpenAI pazi\u0146ojums par GPT-40 n\u0101k tikai vienu dienu pirms Google ikgad\u0113j\u0101s I\/O izstr\u0101d\u0101t\u0101ju konferences, kur\u0101 Google n\u0101ks klaj\u0101 ar saviem ar AI saist\u012btus pazi\u0146ojumus. OpenAI ir ar\u012b apsol\u012bjis, ka gaid\u0101mi vair\u0101ki pazi\u0146ojumi, t\u0101p\u0113c nevar b\u016bt \u0161aubu, ka konkurence starp abiem konkurentiem tikai s\u0101k saasin\u0101t.<\/p>\n

Lai gan m\u0113s nezin\u0101m, kad OpenAI ievies\u012bs vair\u0101k izmai\u0146u sav\u0101 AI model\u012b, m\u0113s zin\u0101m, ka GPT-4o ir paredz\u0113ts ier\u012bc\u0113m, kur\u0101s vis\u0101 pasaul\u0113 darbojas ChatGPT. T\u0101p\u0113c turiet \u012bk\u0161\u0137us un gaidiet, l\u012bdz j\u016bsu ier\u012bc\u0113 tiks par\u0101d\u012bts atjaunin\u0101jums. L\u012bdz n\u0101kamajai reizei!<\/p>\n","protected":false},"excerpt":{"rendered":"

OpenAI nesen pazi\u0146oja par savu GPT-4o modeli\u00a0\u2014 jaunu ChatGPT dzin\u0113ja vado\u0161o modeli\u00a0\u2014 t\u0101 slaveno AI darbin\u0101tu t\u0113rz\u0113\u0161anas robotu, kas pirmo reizi tika izlaists 2022.\u00a0gad\u0101. \u0160is jaun\u0101kais modelis cilv\u0113ka un datora mijiedarb\u012bbu pace\u013c piln\u012bgi jaun\u0101 l\u012bmen\u012b, padarot to \u0101tr\u0101ku un \u0101tr\u0101ku. daudz dabisk\u0101ka. GPT-4o ‘o’ apz\u012bm\u0113 ‘omni’, jo tas apvieno tekstu, att\u0113lu un audio vien\u0101 model\u012b. […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[182,183,130],"class_list":["post-1997","post","type-post","status-publish","format-standard","hentry","category-how-to","tag-artificial-intelligence","tag-chatgpt","tag-microsoft"],"acf":[],"_links":{"self":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/1997","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/comments?post=1997"}],"version-history":[{"count":1,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/1997\/revisions"}],"predecessor-version":[{"id":1998,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/1997\/revisions\/1998"}],"wp:attachment":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/media?parent=1997"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/categories?post=1997"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/tags?post=1997"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}