{"id":5071,"date":"2024-10-28T13:59:47","date_gmt":"2024-10-28T13:59:47","guid":{"rendered":"https:\/\/howtogeek.blog\/lv\/?p=5071"},"modified":"2024-10-28T13:59:47","modified_gmt":"2024-10-28T13:59:47","slug":"claude-35-sonnet-released-by-anthropic-a-new-and-impressive-ai-model","status":"publish","type":"post","link":"https:\/\/howtogeek.blog\/lv\/claude-35-sonnet-released-by-anthropic-a-new-and-impressive-ai-model\/","title":{"rendered":"Kloda 3.5 sonets, ko izlaida Anthropic: jauns un iespaid\u012bgs AI modelis"},"content":{"rendered":"

Nesen Anthropic ir ieviesis uzlabojumus savam vado\u0161ajam modelim Claude 3.5 Sonnet, kas, iesp\u0113jams, ir nok\u013cuvu\u0161i zem radara citu iev\u0113rojamu atjaunin\u0101jumu, piem\u0113ram, datora lieto\u0161anas un anal\u012bzes r\u012bka, d\u0113\u013c. Tom\u0113r atjaunin\u0101tais Claude 3.5 Sonnet ievie\u0161 vair\u0101kus iev\u0113r\u012bbas cien\u012bgus uzlabojumus.<\/p>\n

Lai gan datoru lieto\u0161ana ir noz\u012bm\u012bgs sasniegums, t\u0101 izmanto\u0161ana lietot\u0101ju vid\u016b joproj\u0101m ir ierobe\u017eota. Tas galvenok\u0101rt ir t\u0101p\u0113c, ka tas ir agr\u012bn\u0101 att\u012bst\u012bbas stadij\u0101 un tam tr\u016bkst piln\u012bgas uzticam\u012bbas. Turkl\u0101t t\u0101 pieejam\u012bba ir ierobe\u017eota ar API lieto\u0161anu, padarot to maz\u0101k pieejamu pla\u0161\u0101kai auditorijai.<\/p>\n

Turpretim jaunais Claude 3.5 Sonnets ir pieejams t\u016bl\u012bt\u0113jai lieto\u0161anai un pied\u0101v\u0101 iev\u0113rojami uzlabotu veiktsp\u0113ju. T\u0101tad, kas tie\u0161i at\u0161\u0137ir \u0161o atjaunin\u0101to 3.5 Sonnetu?<\/p>\n

Anthropic zi\u0146ojum\u0101 uzsv\u0113rts, ka jaun\u0101kais Claude 3.5 Sonnet ir uzr\u0101d\u012bjis iev\u0113rojamus uzlabojumus sal\u012bdzin\u0101jum\u0101 ar iepriek\u0161\u0113jiem mode\u013ciem un daudziem konkurentiem nozares etalonos. \u012apa\u0161i iespaid\u012bgi ir sasniegumi a\u0123entu kod\u0113\u0161anas un r\u012bku izmanto\u0161an\u0101. Piem\u0113ram, t\u0101 ir palielin\u0101jusi savu a\u0123entu kod\u0113\u0161anas precizit\u0101ti no 33,4% l\u012bdz 49% SWE stenda p\u0101rbaud\u0113, un a\u0123entu r\u012bku izmanto\u0161anai TAU stenda test\u0101 t\u0101 ir palielin\u0101jusies no 51,0% l\u012bdz 69,2% mazumtirdzniec\u012bb\u0101 un no 22,8% l\u012bdz 46,0% aviosabiedr\u012bbu sektor\u0101. J\u0101atz\u012bm\u0113, ka \u0161ie uzlabojumi tiek sasniegti ar t\u0101d\u0101m pa\u0161\u0101m izmaks\u0101m un \u0101trumu k\u0101 t\u0101 priek\u0161g\u0101j\u0113js.<\/p>\n

\"Claude
Avots: Anthropic<\/figcaption><\/figure>\n

Nov\u0113rt\u0113jot, Claude 3.5 Sonnet (jauns), ko da\u017ei lietot\u0101ji d\u0113v\u0113 par 3.6 Sonnet nosaukumu neskaidr\u012bbas d\u0113\u013c, par\u0101da iev\u0113rojamus sasniegumus, jo \u012bpa\u0161i kod\u0113\u0161anas, argument\u0101cijas un rado\u0161\u0101s rakst\u012b\u0161anas iesp\u0113j\u0101s.<\/p>\n

Iepriek\u0161 atz\u012bts par vienu no lab\u0101kajiem kod\u0113\u0161anas izpild\u012bt\u0101jiem, Klods ir guvis iev\u0113rojamus pan\u0101kumus ar \u0161o atjaunin\u0101jumu, tagad sp\u0113j atrisin\u0101t \u013coti sare\u017e\u0123\u012btas probl\u0113mas ar vienu atbildi. Tas ir ar\u012b uzlabojis sp\u0113ju labot iepriek\u0161\u0113j\u0101s k\u013c\u016bdas.<\/p>\n

Lietot\u0101ji atkl\u0101s, ka jaunais Claude 3.5 Sonnet sniedz atbildes daudz \u0101tr\u0101k, nov\u0113r\u0161ot neapmierin\u0101t\u012bbu, kas radusies ar t\u0101 iepriek\u0161\u0113jo versiju, kuras veiktsp\u0113ja bija iev\u0113rojami pasliktin\u0101jusies.<\/p>\n

Klods ir ar\u012b uzlabojis savas sprie\u0161anas prasmes, sal\u012bdzinot ar OpenAI modeli o1. Lai gan tas var neatbilst o1 sare\u017e\u0123\u012btos sprie\u0161anas uzdevumos, tas joproj\u0101m ir kompetents ar vienk\u0101r\u0161\u0101kiem argument\u0101cijas izaicin\u0101jumiem.<\/p>\n

Lai p\u0101rbaud\u012btu Kloda argument\u0101ciju, m\u0113\u0123iniet to uzdot, How many words are in your response to this prompt?<\/code>. Konkr\u0113ti, transformatoru arhitekt\u016bras mode\u013ci bie\u017ei c\u012bn\u0101s ar \u0161\u0101diem nor\u0101d\u012bjumiem, jo \u200b\u200btiem ir nepiecie\u0161ams paredz\u0113t turpm\u0101kos v\u0101rdus. Tom\u0113r \u0161\u012b uzvedne mudina modeli iepriek\u0161 izstr\u0101d\u0101t strat\u0113\u0123iju.<\/p>\n

OpenAI o1 modelis izce\u013cas, efekt\u012bvi apstr\u0101d\u0101jot \u0161o uzvedni. Lai gan jaunais Claude 3.5 Sonnets nesasniedz o1 prasmes, tas iev\u0113rojami p\u0101rsp\u0113j vec\u0101ku Claude versiju. Iepriek\u0161 Klods, iesp\u0113jams, izdar\u012bja sapr\u0101t\u012bgu min\u0113jumu, bet tagad tas m\u0113\u0123ina prec\u012bzi saskait\u012bt v\u0101rdus, lai gan ne\u0146emot v\u0113r\u0101 tos visus vien\u0101di. Galu gal\u0101 tas, vai Kloda atbilde tiks uzskat\u012bta par pareizu, b\u016bs atkar\u012bgs no person\u012bg\u0101s interpret\u0101cijas.<\/p>\n

Let me count the words in my response to your question:
\nThis (1) is (2) my (3) first (4) response (5) to (6) your (7) prompt (8).
\nThere are 8 words in my response up to this point. Note that I did not count the words after the period since those were explaining my counting process rather than being part of my direct response to your question.<\/code><\/code><\/p>\n

\"Kloda<\/figure>\n

Saska\u0146\u0101 ar sniegto gad\u012bjumu da\u017ei lietot\u0101ji var\u0113tu iebilst, ka Klods prec\u012bzi saskait\u012bja v\u0101rdus, kuros vi\u0146\u0161 uzskat\u012bja savu atbildi. Taj\u0101 ir paskaidrots, ka turpm\u0101kie v\u0101rdi neietilpst t\u0101s atbild\u0113, jo tie vienk\u0101r\u0161i preciz\u0113 argument\u0101cijas procesu. Lai gan da\u017ei var\u0113tu neiev\u0113rot ievadfr\u0101zes \u201c\u013baujiet man saskait\u012bt v\u0101rdus atbild\u0113 uz j\u016bsu jaut\u0101jumu\u201d izlai\u0161anu, citi to var uztvert cit\u0101di. Galu gal\u0101 tas k\u013c\u016bst par person\u012bga sprieduma jaut\u0101jumu.<\/p>\n

Tom\u0113r \u0161is gad\u012bjums par\u0101da, ka jaunais Claude 3.5 Sonnet patie\u0161\u0101m ir uzlabojis savas argument\u0101cijas sp\u0113jas. Tagad tas var formul\u0113t savus dom\u0101\u0161anas procesus vai teikt: “Atgremot par to, pagaidiet”, kad tas prasa papildu laiku, lai apsv\u0113rtu \u2014 nedaudz l\u012bdz\u012bgi k\u0101 o1. Reiz\u0113m tas var pat sevi labot, nor\u0101dot: “\u013baujiet man to p\u0101rdom\u0101t.”<\/p>\n

\u0160\u012b t\u0101lredz\u012bbas sp\u0113ja ir ar\u012b veicin\u0101jusi Kloda rado\u0161\u0101s rakst\u012b\u0161anas prasmes. Ar sp\u0113ju dom\u0101t uz priek\u0161u, tas var izveidot papla\u0161in\u0101tus st\u0101st\u012bjumus, kuros ir sakar\u012bgi loki, priek\u0161v\u0113stures elementi un valdzino\u0161i varo\u0146i.<\/p>\n

Turkl\u0101t ir g\u016bti iev\u0113rojami pan\u0101kumi t\u0101s anal\u012btiskaj\u0101 darb\u012bb\u0101. Claude tagad cie\u0161i konkur\u0113 ar Anthropic pla\u0161\u0101ko modeli Claude 3 Opus, k\u0101 ar\u012b OpenAI o1 mini attiec\u012bb\u0101 uz anal\u012bzi.<\/p>\n

Rezum\u0113jot, jaun\u0101kie atjaunin\u0101jumi ir devu\u0161i iev\u0113rojamu progresu vair\u0101k\u0101s dimensij\u0101s. Jaun\u0101s kod\u0113\u0161anas funkcijas ir izpeln\u012bju\u0161\u0101s lielu uzman\u012bbu. Tom\u0113r pa\u0161reiz\u0113jais Kloda ierobe\u017eojums ir lieto\u0161anas ierobe\u017eojumi, kas ir iev\u0113rojami ierobe\u017eojo\u0161\u0101ki pat Pro lietot\u0101jiem, sal\u012bdzinot ar ChatGPT.<\/p>\n

Avots<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

Nesen Anthropic ir ieviesis uzlabojumus savam vado\u0161ajam modelim Claude 3.5 Sonnet, kas, iesp\u0113jams, ir nok\u013cuvu\u0161i zem radara citu iev\u0113rojamu atjaunin\u0101jumu, piem\u0113ram, datora lieto\u0161anas un anal\u012bzes r\u012bka, d\u0113\u013c. Tom\u0113r atjaunin\u0101tais Claude 3.5 Sonnet ievie\u0161 vair\u0101kus iev\u0113r\u012bbas cien\u012bgus uzlabojumus. Lai gan datoru lieto\u0161ana ir noz\u012bm\u012bgs sasniegums, t\u0101 izmanto\u0161ana lietot\u0101ju vid\u016b joproj\u0101m ir ierobe\u017eota. Tas galvenok\u0101rt ir t\u0101p\u0113c, […]<\/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":[182,245,130],"class_list":["post-5071","post","type-post","status-publish","format-standard","hentry","category-how-to","tag-artificial-intelligence","tag-claude","tag-microsoft"],"acf":[],"_links":{"self":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/5071","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=5071"}],"version-history":[{"count":1,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/5071\/revisions"}],"predecessor-version":[{"id":5072,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/posts\/5071\/revisions\/5072"}],"wp:attachment":[{"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/media?parent=5071"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/categories?post=5071"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/howtogeek.blog\/lv\/wp-json\/wp\/v2\/tags?post=5071"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}