Uyisebenzisa njani i-OpenAI O1? Isikhokelo somsebenzisi, ixabiso kunye nophononongo lwamava

Awukayazi indlela yokusebenzisa i-OpenAI O1 ? Ngokuqinisekileyo "usemva kwexesha"!

Ngeli xesha lokudubula kobukrelekrele bokwenziwa, i-OpenAI O1 izise into entsha yetekhnoloji engazange ibonwe ngaphambili. Ngoku, ndiya kukuthatha ukufumanisa, ukuze uguqule ukusuka "kwi-artificial intelligence blind" ukuya kwi "AI ingcali" ngemizuzwana.

Yintoni i-OpenAI O1?

I-OpenAI O1, igama livakala kakhulu kubuchwephesha obuphezulu, sisizukulwana esitsha sobukrelekrele bokwenziwa kwenkqubo esungulwe yi-OpenAI.

Ukubeka ngokulula, i-O1 "yingqondo ekrelekrele" ekrelekrele kwaye isebenze ngakumbi kunaso nasiphi na isizukulwana sangaphambili setekhnoloji ye-AI. Nokuba kukuhamba komsebenzi okuzenzekelayo okanye ukusetyenzwa kolwimi lwendalo oluntsonkothileyo, i-O1 inokuyiphatha ngokulula.

Ayikongi nje ixesha elininzi kumsebenzi wakho wemihla ngemihla, kodwa ikwaphucula ukuchaneka kwezigqibo kwaye kukunceda wenze ukhetho olukrelekrele.

Ke, nanku umbuzo: Uyisebenzisa njani i-OpenAI O1? Lithini ixabiso? Kuthekani ngamava? Okulandelayo, ndiya kuniphendula nganye nganye.

Uyisebenzisa njani i-OpenAI O1? Isikhokelo sokuqalisa ngokukhawuleza

Unokucinga ukuba ukusebenzisa i-AI kuyinkimbinkimbi, kodwa eneneni, ukusebenza kwe-OpenAI O1 kulula kakhulu, kwaye kunokuthiwa "kukusebenza okufana nobudenge". Landela la manyathelo ambalwa kwaye unokuqalisa ngokulula.

Indlela yoku-1:umrhumoNcokolaGPT Kwaye isebenzisa i-OpenAI O1

Inyathelo lokuqala, ewe, kukubhalisa kwaye ungene kwiakhawunti ye-OpenAI.

• I-ChatGPT Plus kunye nabasebenzisi beQela sele benokufikelela kwi-o1-preview.

Inyathelo lesibini kukubhalisa kwi-ChatGPT Plus.

• Ukubhaliselwa kwe-ChatGPT Plus kubiza i-20 yeedola ngenyanga.
• Oku kuvumela abasebenzisi ukuba babeke phambili i-ChatGPT ngexesha leeyure eziphakamileyo kwaye bafumane iimpendulo ezikhawulezayo.

Ukuba ubhalisa i-OpenAI kwilizwe laseTshayina, ngokukhawuleza "OpenAI's services are not available in your country."▼

Imisebenzi ephucukileyo ifuna ukuba abasebenzisi baphucule ukuya kwi-ChatGPT Plus ngaphambi kokuba isetyenziswe, nangona kunjalo, kumazwe angayixhasi i-OpenAI, kunzima ukwenza i-ChatGPT Plus isebenze, kwaye kufuneka ujongane nemiba enzima efana namakhadi etyala angaphandle.

Apha sikwazisa ngewebhusayithi efikelelekayo kakhulu ebonelela ngeChatGPT Plus ngeeakhawunti ezabelwana ngazo zokurenta.

Nceda ucofe ikhonkco elingezantsi ukuze ubhalise iakhawunti yePlanet ▼

Cofa ikhonkco elingezantsi ukuze ufumane isikhokelo esineenkcukacha zokubhalisa iakhawunti yakho kwiPlanet▼

🔗Ikhonkco lokuthenga i-ChatGPT Plus nge-carpooling kwi-akhawunti yakho yeplanethi, fumana isaphulelo se-50% kwiintlawulo zenyanga!

Iingcebiso:

• Iidilesi ze-IP eRashiya, eChina, eHong Kong, naseMacau azinakubhalisela i-akhawunti ye-OpenAI. Kuyacetyiswa ukuba ubhalise kwenye idilesi ye-IP.

Indlela yesi-2: Sebenzisa i-OpenAI O1 ngokubiza i-API interface

I-OpenAI O1 ineendlela ezahlukeneyo ze-API kwaye inezisombululo ezahlukeneyo kwiimfuno ezahlukeneyo.

• Umzekelo, ukuba ungumphuhlisi, ungakhetha ujongano oluthile lomphuhlisi we-API;
• Ukuba ungumsebenzisi weshishini, i-API yoHlelo loShishino inokukunceda ukuba ulawule kakuhle ukuhamba komsebenzi kwangaphakathi kwenkampani yakho.

 Iiparamitha zoqwalaselo

Musa ukoyika iteknoloji! Nokuba awazi nto malunga netekhnoloji, ujongano lwe-O1 lwenzelwe ukuba lusebenziseke kakhulu. Kufuneka ulandele kuphela i-prompts kwi-interface kwaye ufake iiparameters ozifunayo, ezifana nohlobo lwedatha yolwimi oluza kucutshungulwa, uhlobo lokuvelisa uxwebhu, njl. njl. .

Qhuba umsebenzi kwaye ufumane iziphumo

Xa ufake zonke iiparameters eziyimfuneko, nqakraza "Run". I-O1 iya kuqhubekisa ngokuzenzekelayo imisebenzi oyithumelayo, nokuba ivelisa isicatshulwa, ihlalutya idatha, okanye eminye imisebenzi enzima, kwaye inokugqitywa kwimizuzwana embalwa. Iziphumo ziya kuboniswa ngokuzenzekelayo kwiakhawunti yakho yangasemva, kwaye ungalungisa ngokuqhubekayo okanye ulungiselele njengoko kufuneka.

 Phucula kwaye ulungiselele

Ngaba ucinga ukuba imveliso ye-O1 ayilindelekanga? Sukuba nexhala, i-O1 ixhasa ukwenziwa okuphindaphindiweyo. Unokwenza ngokuthe ngcembe i-AI "ikuqonde" ibe ngcono ngokulungelelanisa iiparamitha okanye ukubonelela ngeesampulu zokufaka ngakumbi kuyo. Okukhona uyisebenzisa ixesha elide, kokukhona umphumo uya kuba ngcono.

Ixabisa malini i-OpenAI O1? Ngaba kufanelekile abantu okanye amashishini?

Uthotho lweemodeli ze-OpenAI ze-o1 luquka iinguqulelo ezimbini: i-o1-preview kunye ne-o1-mini. Ziphuculiwe kubuchule bokuqiqa, ngakumbi kwicandelo lesayensi , itekhnoloji, ubunjineli, kunye nezibalo (STEM). Imodeli ye-o1-preview igqwesile ekuboneleleni ngoluhlu olubanzi "lolwazi lwehlabathi," ngelixa i-o1-mini ingcono kumxholo wokuqiqa ofana nokubhala ikhowudi, kodwa isenokungakwazi kangako ngokubhekiselele kulwimi nolwazi ngokubanzi.

在价格方面，o1-preview模型的API调用价格为每输入100万个token 15美元，每输出100万个token 60美元。相比之下，o1-mini模型的价格较为经济，其费用为每输入100万个token 3美元，每输出100万个token 12美元，这比o1-preview模型便宜了80%。

o1模型的使用限制相对较低，目前允许o1-preview每周使用30次，o1-mini每周使用50次。这些限制可能会随着用户需求和反馈逐步提升。

I-OpenAI iceba ukuqalisa inguqu enkulu yomxholo we-o1 imodeli kwixesha elizayo kwaye ivule ngokuthe ngcembe i-o1-mini kubo bonke abasebenzisi be-ChatGPT bamahhala, nangona ixesha elithile lingekacaciswa. Ngexesha elifanayo, i-OpenAI iphinda isebenze ukunyusa imida yokusetyenziswa ngexesha, kwaye iceba ukongeza imisebenzi efana nokukhangela iwebhu, ifayile kunye nokulayisha imifanekiso, njl., ukwenza imodeli ibe luncedo ngakumbi. Ngokuphathelele kwikamva lamaxabiso emodeli, i-OpenAI yatsho ukuba ngokwembali, amaxabiso ehle amaxesha angama-1 rhoqo kwiminyaka eyi-2-10, kwaye lo mkhwa unokuqhubeka.

Ukuthetha ngexabiso, lo ngumbuzo wonke umntu ochaphazelekayo. Emva koko, nokuba unako "ukuchitha imali encinci ukwenza izinto ezinkulu" ihambelana ngokuthe ngqo nexabiso lale mveliso.

1. Uguqulelo lwesilingo sasimahla

Ukuvumela wonke umntu ukuba abe namava angcono kwi-O1, i-OpenAI inokubonelela ngoguqulelo lwesilingo sasimahla.

Kubasebenzisi abaqhelekileyo, nangona i-o1-mini ingavulwanga okwangoku, kulindeleke ukuba ivuleleke kuye wonke umntu kungekudala.

Nangona imisebenzi ilinganiselwe, kwanele ukuba ube namava okusebenza kwayo okunamandla kunye nojongano olulula lokusebenza. Ukuba uyisebenzisa kuphela ngamaxesha athile, okanye ufuna nje ukufumana ukuqonda kwangaphambili kweempawu ze-O1, le nguqulelo yanele.

2. Uguqulelo olusisiseko

I-ChatGPT Plus kunye nabasebenzisi beQela sele benokufikelela kwi-o1-preview.

Ukubhaliselwa kwe-ChatGPT Plus kubiza i-20 yeedola ngenyanga. Oku kuvumela abasebenzisi ukuba babeke phambili i-ChatGPT ngexesha leeyure eziphakamileyo kwaye bafumane iimpendulo ezikhawulezayo.

Le nguqulo ifanelekile kumashishini amancinci kunye nabasebenzisi ngabanye, kwaye imisebenzi yayo igubungela ezininzi iimfuno zemihla ngemihla, njengokuvelisa umbhalo, uhlalutyo lwedatha, njl. Inzuzo yayo ilele ekuguquguqukeni kwayo okuphezulu kwaye ifanelekile kubasebenzisi abafuna ukusetyenziswa rhoqo kodwa banohlahlo lwabiwo-mali olulinganiselwe.

3. Ushicilelo lweShishini

Inguqulo yeshishini yenzelwe iinkampani ezinkulu kunye neziphakathi Ixabiso liphezulu kakhulu, kodwa imisebenzi inamandla kakhulu. Ayinakuphatha kuphela inani elikhulu ledatha, inokuphinda yandise ngokupheleleyo kwaye iphucule ukuhamba komsebenzi kwenkampani. Ukuba uyinkampani efuna ukuphucula imveliso kunye nokwandisa amandla okusebenza kwedatha, le nguqulo ngokuqinisekileyo kufanelekile ukutyala imali kuyo.

Anjani amava akho usebenzisa i-OpenAI O1? Uphononongo lulapha!

Njengomsebenzisi wamava anzulu, kufuneka nditsho ukuba ukusebenza kwe-OpenAI O1 ngokwenene "kuyamangalisa".

Inobulumko kwaye ikhawuleza kunayo nayiphi na inkqubo ye-AI yangaphambili malunga nokuveliswa kombhalo okrelekrele, ukuguqulelwa kunye nohlalutyo lwedatha.

1. Isantya kunye nokusebenza kakuhle

Ngokumalunga nesantya, ixesha lokuphendula le-O1 lifutshane kakhulu. Nokuba yeyiphi na imisebenzi oyingenisayo, inokugqityezelwa ngemizuzwana. Ngokukodwa kwimisebenzi efuna ukusetyenzwa kweedatha ezinkulu, ukusebenza kwe-O1 kubalaseleyo. Xa kuthelekiswa nezinye iinkqubo ze-AI, ubuncinane i-30% isebenza kakuhle.

2. Ukuchaneka kunye nobukrelekrele

I-O1 ayikhawulezi nje, kodwa "ichanekile". Iyaziqonda ngokuchanekileyo iimfuno zabasebenzisi kwaye ivelise ngokuzenzekelayo umxholo wemveliso ephezulu. Ngokumalunga nokusetyenzwa kolwimi lwendalo, i-O1 inokuchonga ngokuzenzekelayo iimeko ezahlukeneyo kwaye ivelise isicatshulwa esifanelekileyo okanye iziphumo zoguqulo ngokusekelwe kumxholo.

3. Ujongano lomsebenzisi

Uyilo lojongano lwe-O1 lulula kwaye lucacile, nto leyo eyenza kube lula nakubantu abangenamvelaphi yobuchwephesha ukuba baqalise. Kukho izikhokelo ezicacileyo kunye namaxwebhu oncedo kwisinyathelo ngasinye, ngoko akukho miqobo kwinkqubo yokusetyenziswa.

4. Ukubaleka

Kubasebenzisi beshishini, eyona nto ibalaseleyo ye-O1 kukulinganisa kwayo. Unokuphucula imisebenzi kunye nokwandisa i-API nangaliphi na ixesha ngokweemfuno zoshishino. Nokuba iimfuno zakho zitshintshile, i-O1 inokuziqhelanisa ngokukhawuleza, ikugcina uphambili kukhuphiswano.

o1 series version intshayelelo

Olu luhlu lwe-o1 luneenguqulelo ezimbini: i-o1-mini kunye ne -o1-preview.

o1-umboniso

• Lo ngumboniso wemodeli ye-o1 esemthethweni ezayo.
• I-o1 izuze impumelelo ebalulekileyo ekuqiqweni kobukrelekrele bokwenziwa.

o1-mini

• Le yimodeli yentelekelelo ekhawulezayo nefikelelekayo eqhuba kakuhle ngokukodwa kwimisebenzi yokucwangcisa.
• Njengoguqulelo oluncinci, i-o1-mini inexabiso kuphela kwi-o1-preview 20%, ngelixa ugcina ukuqiqa okusebenzayo kwaye ukhula ngakumbi kwaye ufikeleleka.

Indlela amandla okuqiqa e-OpenAI O1 asebenza ngayo ngokwenene

I-OpenAI icacisa ngokukodwa ukuba ezi modeli zintsha, eziqeqeshwe ngokufunda okuqinisiweyo , ziyakwazi ukusingatha imisebenzi enzima yokuqiqa.

Ngoko ke, kwimeko yomzekelo omkhulu wolwimi, kuthetha ukuthini kanye kanye "amandla okuqiqa" ?

Njengokuba abantu kufuneka bacinge nzulu xa bejongene neengxaki ezinzima, u-O1 usebenzisa "uthotho lweengcinga" xa esombulula iingxaki.

• Icotha kancinane amanyathelo antsonkothileyo abe ngamacandelo alawulekayo ngokuchonga nokulungisa iimpazamo;
• Ukuba isicwangciso sangoku siyasilela, siya kuzama iingcamango ezintsha.

Inkqubo yokuqiqa ibandakanya la manyathelo alandelayo:

1. Yenza iithokheni zentelekelelo;
2. Iziphumo ezibonakalayo zophawu lwempendulo;
3. Susa iziphawuli zentelekelelo kumxholo,

Gcina umxholo ugxile kulwazi oluphambili ngokususa abamakishi be-inference.

Kubalulekile ukuqaphela ukuba nangona abamakishi be-inference bengabonakali kwi-API, basahlala kwindawo yefestile yemodeli yomxholo kwaye babandakanyiwe kwiindleko zabamakishi bemveliso.

Imibono kaJim Fan

Umphandi ophezulu we-NVIDIA uJim Fan ubonise ukuba nangona le nkqubo yokuqikelela isenokuba icotha, indlela esetyenziswa ngayo ixesha elide lokuqikelela iye yathandwa kakhulu kwiindawo zemveliso.

UJim uphinde wabonelela ngeembono ezininzi ezinomdla:

1. Ukuqiqa okukrelekrele akuxhomekekanga kwiimodeli ezinkulu : iiparameter zeemodeli ezininzi ezinkulu zisetyenziselwa ukugcina ulwazi oluyinyani ukujongana neengxaki zesiqhelo. Nangona kunjalo, singahlula amandla okuqiqa kwiziseko zolwazi. Khawucinge "ngesiseko sokuqiqa" esinobuchule esinokukwazi ukubiza izixhobo ezahlukeneyo (ezifana nokukhangela kwiwebhu okanye izixhobo zokuhlola ikhowudi) ukuze sifumane ulwazi oluyimfuneko. Icebo elinjalo linokunciphisa kakhulu izixhobo zokubala ezifunekayo kuqeqesho lwe-AI.

2. Le modeli intsha itshintsha umsebenzi omkhulu wokubala ukusuka kwisigaba soqeqesho ukuya kwisigaba sokusetyenziswa okusebenzayo. Unokucinga ngemodeli enkulu yolwimi njenge "hlabathi eliqikelelweyo" elisekelwe kwisicatshulwa. Xa usombulula ingxaki, le modeli izama iindlela ezahlukeneyo kunye neemeko kweli "hlabathi eliqikelelweyo" ukufumana isisombululo esifanelekileyo. Le nkqubo ifana kancinci nesoftware yokudlala i-chess ( njenge -AlphaGo) eqikelelweyo iintshukumo ezininzi engqondweni yayo ukuze imisele intshukumo elandelayo engcono.

Ukuthelekiswa kwe-o1 kunye ne-GPT-4o

Ukuvavanya ukusebenza kwemodeli ye-o1 xa kuthelekiswa ne -GPT-4o , i-OpenAI yenze uvavanyo olubanzi kunye nemilinganiselo yokufunda koomatshini.

Iziphumo zibonisa ngokucacileyo ukuba i-o1 iphumelela kakhulu kwi-GPT-4o kwimisebenzi enzima yokuqiqa efana nezibalo, ucwangciso, kunye neengxaki zesayensi .

Ngokukodwa, i-o1 iqhube kakuhle kakhulu kuvavanyo lokulinganisa lwe-GPQA-diamond . Olu vavanyo lwenzelwe ukuvavanya ubuchule bemodeli kwicandelo lekhemistri, ifiziksi, kunye nebhayoloji .

Ukuthelekisa ukusebenza kwemodeli nabantu, i-OpenAI imeme abanini be-PhD ukuba baphendule imibuzo efanayo ye-GPQA-diamond.

Iziphumo ezothusayo

I-o1 yodlula ezi ngcaphephe, yaba yimodeli yokuqala ye-AI ukugqwesa iiPhDs kolu phawu. Nangona oku akuthethi ukuba i-O1 ingcono kune-PhD kuzo zonke iindawo, ibonisa ukuba inobuchule ngakumbi ekusombululeni iingxaki ezithile.

Ungayibona ingxelo yobugcisa kwimodeli ye-O1 apha .

Umzekelo wokuthelekisa: o1 kunye ne-GPT-4o

Masithelekise ukusebenza kwe-o1 kunye nemodeli yangaphambili ye-GPT-4o ngombuzo oqhelekileyo: Zingaphi ii-"r" ezikwigama elithi "strawberry"?

Prompt: How many ‘r’ letter are in the word strawberry?

• o1 echithiweyoImizuzwana eli-33, esetyenzisiweyo296 amanqaku, wanikela impendulo echanileyo;
• I-GPT-4o kuphela ayisebenzisiImizuzwana eli-1, ukusetyenziswa39 amanqaku, kodwa akakwazanga ukuphendula ngokuchanileyo.

Zama omnye umbuzo:

Zombini ezi modeli zacelwa ukuba zidwelise amazwe amahlanu amagama awo anonobumba "A" kwindawo yesithathu.

Prompt: Give me 5 countries with letter A in the third position in the name

• I-o1 iphinde yanika impendulo echanekileyo, nangona kuthathe ixesha elide "ukucinga" kune-GPT-4o.

Iingcinga zam kwi-OpenAI O1: isixhobo sotshintsho okanye i-gimmick?

Isishwankathelo, i-OpenAI O1 ngokuqinisekileyo "sisixhobo soguqulo". Ayenzi lula kakhulu ukusetyenzwa kwemisebenzi entsonkothileyo, kodwa yenza ukuba ukusetyenziswa kwe-AI kudume ngakumbi kwaye kusebenze. Kubasebenzisi ngabanye, i-O1 ingumncedi olungileyo wokuphucula ukusebenza kakuhle kunye nokunciphisa umthwalo womsebenzi kubasebenzisi beshishini, izise inguqu yemveliso engazange ibonwe ngaphambili.

Nangona kunjalo, kufuneka kuqatshelwe ukuba i-O1 ayinamandla onke. Nangona umsebenzi wayo unamandla kakhulu, ikwafuna iindleko ezithile zokufunda kunye nenkqubo yokulungelelanisa. Ukuba eli lixesha lakho lokuqala usebenzisa isixhobo se-AI, kunokuthatha ixesha ukuqonda ingqiqo nokusebenza kwayo.

Isishwankathelo: Kutheni ufuna i-OpenAI O1?

I-OpenAI O1 ingaphezulu nje kwesixhobo se-AI, lelona qabane libalaseleyo lokuphucula ukusebenza kakuhle kunye nokwandisa ishishini. Nokuba ungumsebenzisi ozimeleyo okanye umenzi wezigqibo wenkampani, i-O1 inokukunceda ukuba ufumane iziphumo kabini ngesiqingatha somzamo kumsebenzi wakho wemihla ngemihla.

Ke ngoko, ukuba usathandabuza, unokuthatha ithuba lokuzama i-OpenAI O1 kwaye ube namava oguqulo lwamandla oluwazisayo. Khumbula, kweli xesha lokuqhuma kolwazi, kuphela abo "basebenzisa kakuhle i-AI" abanokuthi bangoyiswa.

Yiya ujonge i-OpenAI O1 kwaye uqale ixesha elitsha lomsebenzi osebenzayo kuwe!