Thola amathuluzi, izindaba, nezifundo
Indawo eyodwa yokuhlola konke okuqukethwe kwe-AI okushicilelwe.
“AI”Imiphumela engu-936 kulesi sigaba
Amathuluzi e-AI
936
Ukucubungula izithombe nge-AIIsilinganiso 8Erase.bgI-Erase.bg iyithuluzi eliku-inthanethi elisebenzisa ubuhlakani bokwenziwa ukucubungula ingemuva lezithombe. Lisetshenziswa kakhulu ukubona into esemqoka bese lisusa ingemuva ezithombeni zabantu, ezifuywayo nezimpahla. Ngemva kokulayisha isithombe, ithuluzi lenza ukusika ngokuzenzakalelayo, likhiphe isithombe esinengemuva elisobala, futhi livumele nokusethwa kwengemuva elinombala owodwa. Lifanele abathengisi bezohwebo lwe-inthanethi, abaklami, abadali bokuqukethwe kanye nabasebenzisi abafuna ukuhlela izinto zezithombe. Amabhizinisi angalihlanganisa nokucubungula izithombe ngobuningi noma nokuxhumeka nge-API lapho edinga ukucubungula izithombe eziningi. Imiphumela yangempela ingathinteka imiphetho yento, izinwele noma uboya, izinto ezisobala, kanye nomehluko phakathi kwento esemqoka nengemuva; izithombe eziyinkimbinkimbi zingase zisadinga ukuhlolwa ngesandla noma ukulungiswa okwengeziwe.
Amathuluzi omsindo e-AIIsilinganiso 8SoundfulI-Soundful iyithuluzi lomsindo elisebenzisa ubuhlakani bokwenziwa, elihlanganisa ukukhiqizwa komculo nomtapo womculo ukuze linikeze abasebenzisi izinto zomculo ongemuva. Abasebenzisi bangakhetha uhlobo lomculo, isigqi, umuzwa noma indikimba kuqala, bese bekhiqiza izingoma futhi bazilande noma baqhubeke nokuzihlela ngokwezidingo zabo; le nkundla iphinde ibe nomculo owenziwe ngobuhlakani bokwenziwa owabiwe abadali bomculo. Okuqukethwe kuhlanganisa izitayela ezifana ne-EDM, Deep House ne-Hip Hop, futhi kulungele abadali bokuqukethwe, abakhiqizi bomculo, amabhrendi namabhizinisi kanye nabathandi abazimele, ukuze kusetshenziswe kumavidiyo, amaphodikasti, ukusakaza bukhoma, izikhangiso, imidlalo noma imisebenzi yomuntu siqu. Ezinye izinhlelo zisekela ukulandwa komsindo nge-MP3, WAV nezinye, kanye namakhono okuqhubeka nokukhiqiza afana namathrekhi ahlukanisiwe e-STEM namafayela e-MIDI. Le nkundla inikeza izindlela zamahhala nezikhokhelwayo; izimvume zokulanda nezici ezitholakalayo zingahluka kuye ngohlelo, ngakho kufanele kuqinisekiswe imigomo yelayisensi efanele ngaphambi kokusetshenziswa okusemthethweni kwezohwebo.
Amathuluzi Okuhlela nge-AIIsilinganiso 7.4GreptileIGreptile iyithuluzi lokubuyekeza ikhodi elisebenzisa ubuhlakani bokwenziwa, elenzelwe inqubo yokuthuthukiswa kwesoftware. Ligxila ekuhlaziyeni umongo wendawo yokugcina ikhodi, futhi linikeza imibono yokubuyekeza mayelana namaphutha angaba khona, izinkinga zokuphepha namaphethini angalungile ezicelweni zokuhlanganisa izinguquko. Lingafakwa ezinqubweni zeGitHub neGitLab, ukuze amaqembu achaze izindinganiso zokuhlela ngolimi olujwayelekile, bese esebenzisa le mithetho ekuhloleni izinguquko. Umkhiqizo ulungele amaqembu okuthuthukisa, abaholi bezobuchwepheshe nabaphathi bamaphrojekthi omthombo ovulekile abadinga ukugcina izindawo zokugcina amakhodi eziyinkimbinkimbi, ukuhlanganisa izindinganiso zeminikelo noma ukunciphisa umthwalo wokubuyekeza okwenziwa ngesandla. Ekusetshenzisweni kwangempela, imiphumela yokubuyekeza kusadingeka iqinisekiswe onjiniyela ngokuhambisana nomongo wephrojekthi, futhi ayikwazi ukuthatha indawo yokwahlulela komuntu; ukumbozwa kwezilimi nokusebenza kokubona umqondo webhizinisi oyinkimbinkimbi nakho kufanele kuqinisekiswe endaweni ethile yokugcina ikhodi.
Amathuluzi ehhovisi e-AIIsilinganiso 7.8QuadraticI-Quadratic iyindawo yokusebenzela yedatha eyenzelwe abahlaziyi bedatha namaqembu, ehlanganisa ukusebenza ngamaspredishithi, ukubuza ngemibuzo yolimi lwemvelo, kanye nekhodi yePython ne-SQL endaweni eyodwa yokuhlaziya. Abasebenzisi bangangenisa noma baxhumanise idatha, benze imibuzo, ukuhlanza idatha, izibalo nokuboniswa kwedatha kumaseli, futhi balondoloze izinyathelo zokuhlaziya ukuze kube lula ukuzihlola, ukuzilungisa nokwabelana ngazo. Ilungele ochwepheshe bomkhiqizo, bezokumaketha, bezezimali, bezokusebenza, besayensi yedatha kanye nabemfundo nocwaningo ekuphatheni idatha ekumatafuleni amaningi nemisebenzi yokuhlaziya ephindaphindwayo, futhi isekela nokusebenzisana kwamaqembu. Umkhiqizo unikeza izindlela zamahhala nezikhokhelwayo. Kubalulekile ukuqaphela ukuthi amafomula, imibuzo noma ikhodi ekhiqizwa ngolimi lwemvelo kusadingeka ihlolwe ngokuhambisana nomqondo webhizinisi nedatha yoqobo; imisebenzi eyinkimbinkimbi nayo idinga isisekelo esifanele sokuhlaziya idatha.
Ukucubungula izithombe nge-AIIsilinganiso 8.2ShrinkI-Shrink iyithuluzi lokucindezela izithombe eliku-inthanethi, elisetshenziswa kakhulu ukulinganisa usayizi wefayela kanye nekhwalithi ebonakalayo. Abasebenzisi bangalayisha izithombe ngokuqondile kukhompyutha yabo noma banamathisele i-URL yesithombe ukuze basisebenze, ngaphandle kokubhalisa. Ithuluzi lisekela amafomethi afana ne-PNG, JPG, JPEG, WebP ne-HEIC, futhi livumela ukulungiswa kwekhwalithi nosayizi wesithombe. Lifanele abathuthukisi bewebhusayithi, abadali bokuqukethwe, abathuthukisi bezinhlelo zokusebenza zeselula, abathengisi be-e-commerce kanye nabasebenzisi ngabanye. Izithombe ezicindezelwe zingasetshenziswa kumawebhusayithi, ezinkundleni zokuxhumana, kuma-imeyili noma ekulayishweni kwamafomu. Qaphela ukuthi ukucindezela kuvame ukubangela ushintsho oluthile kwikhwalithi yesithombe, ngakho izithombe ezibalulekile kufanele zihlolwe kuqala ngemva kokucindezelwa; izidingo zokucubungula ngobuningi kanye ne-API azizona izindlela eziyinhloko zokusebenzisa leli thuluzi lewebhu lamahhala okwamanje.
Amathuluzi omsindo e-AIIsilinganiso 7.8NarakeetI-Narakeet iyithuluzi lomsindo eliku-inthanethi elisebenzisa ubuhlakani bokwenziwa ukuguqula umbhalo, amadokhumenti noma amaslayidi abe amavidiyo nomsindo onokulandisa. Abasebenzisi bangafaka umbhalo ngokuqondile noma balayishe amafayela e-PowerPoint, Google Slides namanye, bese uhlelo lukhipha okuqukethwe lukhiqize ividiyo enokulandisa, kunciphise umsebenzi wokuqopha nowokuvumelanisa ngesandla. Umkhiqizo usekela izilimi eziyi-100 namazwi obuhlakani bokwenziwa angaphezu kuka-900, futhi uqondiswe kothisha, abaqeqeshi, abakhangisi, abadali bokuqukethwe ezinkundleni ezizimele kanye namaqembu onjiniyela. Ungaxhunywa nasezinhlelweni ezizenzakalelayo nge-API. Ulwazi olutholakalayo lubonisa ukuthi kunendlela yamahhala nekhokhelwayo; lapho usetshenziswa, kusafanele kuhlolwe ukuphimisela, iphimbo, umugqa wesikhathi nomphumela wokugcina.
Ibhokisi lamathuluzi e-AIIsilinganiso 8LMArenaI-LMArena iyinkundla yokuhlola amamodeli obuhlakani bokwenziwa eqhutshwa umphakathi, eholwa ithimba lezemfundo lase-University of California, Berkeley. Inqubo yayo eyinhloko iwukuthi amamodeli amabili anamagama afihliwe aphendule umbuzo ofanayo, bese abasebenzisi bevota ngokusekelwe kwikhwalithi yezimpendulo; ubunikazi bamamodeli buboniswa ngemva kokuvota. Inkundla isebenzisa imiphumela yamavoti kanye nohlelo lwamaphuzu i-Elo olufana nolusetshenziswa ku-chess ukuze yakhe izikhundla eziguqukayo, futhi ihlanganisa imisebenzi yombhalo, izithombe neminye. Ilungele abacwaningi bobuhlakani bokwenziwa, onjiniyela, abathanda ubuchwepheshe kanye nabasebenzisi abajwayelekile abafuna ukuqhathanisa amamodeli ngokuvundlile. Iphinde inikeze ngamasethi edatha avulekile azosetshenziswa ocwaningweni. Kumele kuqashelwe ukuthi izikhundla zibonisa okuthandwa ngabasebenzisi bangempela kanye nokusebenza emisebenzini ethile, futhi azikwazi ukuthatha indawo yokuhlolwa okukhethekile okuqondene nezidingo zebhizinisi, izindleko noma ukuzinza.
Amathuluzi Okudizayina we-AIIsilinganiso 8RoomsGPTI-RoomsGPT iyithuluzi lokuklama elisebenzisa ubuhlakani bokwenziwa, ikakhulukazi ngokulayisha izithombe zangempela zezindawo zangaphakathi noma zangaphandle ukuze likhiqize izithombe zemiklamo yokuhlobisa ngezindlela ezihlukahlukene. Abasebenzisi bangasebenzisa imiphumela ekhiqiziwe njengesethenjwa esibonakalayo sokuguqula indawo nokuhlanganisa umhlobiso, ngezindlela ezifana nesimanje, ubulula obukhulu kanye nesitayela sakudala. Lifanele abanikazi bezindlu abalungisayo, abaklami bangaphakathi, abadayisi bezindlu kanye nabaqashi, ukuze bahlele imibono yokuklama, benze iziphakamiso noma babonise imibono yokuguqula izindlu. Umkhiqizo uhlinzeka ngezindlela zokusebenzisa zamahhala nezikhokhelwayo, futhi usekela ukusetshenziswa kumadivayisi eselula. Izithombe ezikhiqiziwe zimelela izixazululo ezibonakalayo futhi azikwazi ukuthatha indawo yokulinganisa indawo, imidwebo yokwakha noma isabelomali sangempela; ngaphambi kokusebenzisa umklamo, kusadingeka kuqinisekiswe ubukhulu bendawo, izinto zokwakha kanye nezimo zokwakha.
Amathuluzi Okuthengisa e-AIIsilinganiso 7.8Word SpinnerI-Word Spinner iyinkundla eku-inthanethi yokuguqula umbhalo ngobuhlakani bokwenziwa, eyenzelwe ukudala okuqukethwe nezimo zokumaketha. Isetshenziswa ikakhulu ukulungisa indlela yokubhala, ithoni nokufundeka kombhalo okhiqizwe ubuhlakani bokwenziwa. Abasebenzisi bangafaka noma bakhiqize umbhalo, bese bekhetha ulimi, isitayela sokubhala kanye nezilungiselelo ezifanele ukuze bawuguqule. Bangasebenzisa nokwenza kahle izinjini zokusesha, ukucubungula ngobuningi nezifanekiso ezisethwe ngaphambilini ekuhleleni okuqukethwe. Le nkundla iqondiswe kubakhangisi bokuqukethwe, ochwepheshe bokwenza kahle izinjini zokusesha, amaqembu ezinkampani kanye nabasebenzisi abadinga usizo lokubhala; ulwazi luthi nabafundi bangayisebenzisa. Ikhasi layo lithi inganciphisa izimpawu zokutholwa kombhalo okhiqizwe ubuhlakani bokwenziwa, kodwa imiphumela yokuguqula umbhalo isadinga ukubuyekezwa umuntu futhi ayiqinisekisi imiphumela yokutholwa, ukunemba kwamaqiniso noma ukufaneleka kokuqukethwe.
Ibhokisi lamathuluzi e-AIIsilinganiso 8.5PoeI-Poe iyinkundla ehlanganisa amamodeli obuhlakani bokwenziwa eyakhiwe yi-Quora, egxile ekuzameni nasekuqhathaniseni amamodeli amaningi obuhlakani bokwenziwa kusevisi eyodwa. Abasebenzisi bangashintsha phakathi kwamamodeli ahlukene, benze izingxoxo noma bakhiqize okuqukethwe, futhi badale, balungiselele futhi babelane ngamarobhothi enzelwe izinhloso ezithile. Inkundla iyatholakala kuwebhu nakumadivayisi eselula, futhi inikeza amakhono ahlobene nombhalo, izithombe, amavidiyo nomsindo. Ilungele abathandi bezobuchwepheshe, abathuthukisi, ababhali, abakhangisi nabafundi. I-Poe inikeza izindlela zokusebenzisa zamahhala nezikhokhelwayo; amamodeli nezici ezitholakalayo kufanele kuqinisekiswe ekhasini lamanje lenkundla. Amamodeli ahlukene anekhwalithi yokukhiphayo nemisebenzi afaneleka kuyo engafani, ngakho imiphumela ekhiqiziwe kusafanele ihlolwe umuntu.
Ukukhiqiza ividiyo nge-AIMiniMax H3I-MiniMax H3 iyimodeli ejwayelekile yokukhiqiza esebenza ngazo zonke izinhlobo zemidiya, eyenzelwe ukweqa imingcele yemisebenzi ethile futhi isuke ekukhiqizeni amavidiyo kuphela iye ekukhiqizeni okuqukethwe okuphelele. Inamandla okuqonda umongo wemidiya eminingi, ihlanganise imithombo yolwazi eyinkimbinkimbi efana nezithombe, umsindo namavidiyo, futhi isekele ukwenziwa kokwethulwa okunembile kombhalo nomkhiqizo, ukudluliswa kokunyakaza kwe-V2V kanye nokuhlelwa kokuqukethwe kwemidiya eminingi ngemiyalelo yolimi lwemvelo. Le modeli yenzelwe abadali abasebenza ezimeni zezentengiso ezifana nokukhangisa, ezohwebo lwe-inthanethi nemidlalo, ibanikeze amandla okukhiqiza ezinga lezentengiso ezimweni eziningi futhi ibasize benze izinhloso zokudala eziyinkimbinkimbi ngempumelelo.
Ukuhlela nge-AISALMONN-2 – imodeli yolimi olukhulu yomsindo ejwayelekile ekhishwe yi-Tsinghua nabanyeI-SALMONN-2 – imodeli yolimi olukhulu yomsindo ejwayelekile ekhishwe yi-Tsinghua nabanye iyimodeli yolimi olukhulu yomsindo ejwayelekile eyakhiwe kusetshenziswa i-SPEAR unified self-supervised audio encoder yakwa-ByteDance kanye ne-multi-layer feature fusion adapter, ne-Qwen3 njengesisekelo sombhalo. Isekela imisebenzi ehlukahlukene efana nokuqaphela inkulumo, ukuchaza umsindo, ukuqonda umculo, ukuqaphela imizwa nokuqinisekisa isikhulumi, futhi ngokokuqala ngqa ku-ALLM ejwayelekile ihlanganisa ngokuhlelekile amakhono athuthukile okuhlonza izehlakalo zomsindo, ukuthola umsindo okhohlisiwe, ukuhlola ikhwalithi yenkulumo kanye nokuqaphela inkulumo okunomongo we-multimodal. Imodeli isebenzisa i-SPEAR encoder eyodwa esikhundleni sohlelo oluvamile lwama-encoder amabili, futhi ihlanganisa izici zomsindo zamazinga amaningi nge-MLF adapter ukuze ithole ukusebenza okulinganiselayo ezizindeni ezahlukene. Ngokuqeqeshwa okuphumelelayo kusetshenziswa cishe amahora angu-18,000 edatha egadiwe kuphela, i-SALMONN-2 ifinyelela ukusebenza okuhamba phambili phakathi kwamamodeli omthombo ovulekile anosayizi ofanayo kuma-benchmark amathathu aphelele i-MMAU-Pro, i-MMAR kanye ne-MMSU. Ilungele izimo ezifana nabasizi bemihlangano abahlakaniphile, ukuhlolwa kokuqukethwe komsindo kanye nokuqapha ikhwalithi yenkulumo.
Ukuqaphela inkulumo nge-AII-Hy ASR 3.0 preview – imodeli yesizukulwane esisha yokuqaphela inkulumo eyethulwe yi-Tencent HunyuanI-Hy ASR 3.0 preview – imodeli yesizukulwane esisha yokuqaphela inkulumo eyethulwe yi-Tencent Hunyuan iwuhlelo olusha lokuqaphela inkulumo olwakhelwe phezu kwemodeli yolimi enkulu i-Hy3, oluhlanganisa ukuqaphela inkulumo okunembe kakhulu nokuqonda okujulile kwencazelo. Isekela isiShayina, isiNgisi, isiCantonese kanye nezifunda ezingu-10 zezilimi zesigodi, futhi inamandla okulungisa amaphutha ngokuzenzakalelayo ngokusekelwe kumongo, ukufaka amagama abalulekile, kanye nokusebenza okuqinile ezimweni eziyinkimbinkimbi ezinjengomsindo omkhulu nokuhleba. Kuqoqo lokuhlola elivulekile, i-WER igcinwa cishe ku-3%, kanti iqoqo lokuhlola elizakhelwe ngaphakathi lihola izimbangi kuzo zonke izici. Le modeli ifanele izimo ezifana nesevisi yamakhasimende ehlakaniphile, ukudala okuqukethwe, ukusesha ngezwi, ukusebenzisana ehhovisi namadivayisi ahlakaniphile. Ingaxhunywa nge-Tencent Cloud API noma ihlolwe mahhala ku-Tencent Yuanbao.
Amathuluzi ehhovisi e-AIIsilinganiso 8.3我来我来 (wolai) iyithuluzi lehhovisi elinikwa amandla yi-AI, elenzelwe abantu ngabanye namaqembu, elihlanganisa ukuphathwa kolwazi, ukugcinwa kolwazi nokusebenzisana ku-inthanethi. Abasebenzisi bangahlanganisa umbhalo, izithombe, amathebula nezinto okufanele zenziwe kusihleli esisekelwe emabhulokhini, bese besebenzisa izixhumanisi eziya nhlangothi zombili ukwakha ukuxhumana phakathi kwamakhasi. Amaqembu wona angahlela ndawonye emakhasini abiwe, aphathe imibhalo yamaphrojekthi nemisebenzi. Lilungele ukuphathwa kolwazi, amanothi ezifundo, ukuhlela izinto zocwaningo, ukubhala imibhalo emide kanye nezinqolobane zolwazi zamaqembu. Linikeza izindlela zamahhala nezikhokhelwayo. Indlela yalo yokuhlela iyaguquguquka, ngakho abasebenzisi kudingeka bazihlelele ngokwabo ukwakheka kwamakhasi nobudlelwano bezixhumanisi; ukugcinwa kwesikhathi eside kokuqukethwe okuyinkimbinkimbi nakho kudinga ukuhlelwa okuqhubekayo.
Amathuluzi Okuthengisa e-AIIsilinganiso 8.2CleanlistI-Cleanlist iyinkundla yokucebisa idatha ye-B2B eyenzelwe amaqembu e-GTM. Isekela ukusesha izinkampani noxhumana nabo ngolimi lwemvelo, ukuqinisekisa ama-imeyili nezinombolo zocingo, ukucubungula uhlu lwe-CSV, nokuvumelanisa imiphumela ne-CRM noma amathuluzi okuthengisa. Umkhiqizo uhlinzeka ngohlelo lokusebenza lwewebhu, isandiso se-Chrome, i-REST API, kanye neseva ye-MCP yamaklayenti e-AI afana no-Claude.
Amathuluzi Okuthengisa e-AIIsilinganiso 7.7REVERI-REVER iyithuluzi le-SaaS lokuphatha ukubuyiswa kwemikhiqizo elenzelwe i-e-commerce ewela imingcele nezitolo ezizimele zamabhrendi. Ligxile ekuhlanganiseni ukubuyiswa kwemikhiqizo, ukubuyiselwa kwemali kanye nokushintshwa kwemikhiqizo kube ulwazi olulodwa lwenkonzo yangemva kokuthenga. Abathengisi bangahlela izinyathelo zokufaka izicelo, ukuzibuyekeza nokuzamukela, ukubuyisa imikhiqizo ngokuthunyelwa kanye nokubuyisela imali ngokusekelwe kunqubomgomo yabo yokubuyisa. Bangasebenzisa nokushintsha imikhiqizo, amakhadi esipho noma amaphuzu esitolo ukuze banciphise ukulahleka kwama-oda. Lo mkhiqizo ulungele amabhrendi athengisa izigaba ezifana nezingubo nezicathulo, ezivame ukuba namazinga aphezulu okubuyiswa kwemikhiqizo, kanye namaqembu okusebenza adinga ukuphatha ukuthuthwa okubuyela emuva emazweni amaningi. Imisebenzi yawo ihlanganisa ukuhlaziywa kwedatha yokubuyiswa kwemikhiqizo, izilimi eziningi, izimali eziningi kanye nezimo lapho umkhiqizo uthengwa ku-inthanethi kodwa ubuyiselwa esitolo. Ububanzi bokusetshenziswa obutholakalayo buncike ezinkundleni ezisetshenziswa umthengisi, eziteshini zokuthumela kanye nasekulungiselelweni kwezinqubomgomo.
Amathuluzi okudweba nge-AIIsilinganiso 7.6UltralyticsI-Ultralytics iyinkundla yamathuluzi yokuthuthukisa umbono wekhompyutha, ehlinzeka ngokugeleza komsebenzi kusukela ekuphathweni kwedatha, ekuqeqeshweni kwamamodeli kuya ekusetshenzisweni kwawo. Okuqukethwe okuyinhloko kuhlanganisa amamodeli asukela ku-YOLOv5 kuya ku-YOLO11 kanye nenkundla yokuqeqesha ebonakalayo i-Ultralytics HUB. Abasebenzisi bangakhetha imisebenzi efana nokutholwa kwezinto, ukuhlukaniswa kwezithombe, ukuhlukaniswa kwezingxenye zesithombe nokulinganiswa kokuma, balayishe idatha ukuze baqeqeshe, bese bekhipha amamodeli abe amafomethi afana ne-ONNX, TensorFlow noma i-CoreML ukuze asetshenziswe kumafoni nakumadivayisi asemaphethelweni enethiwekhi. Ilungele abathuthukisi, abacwaningi namabhizinisi adinga ukufaka imisebenzi yombono ezixazululweni zangempela. Kubalulekile ukuqaphela ukuthi ngemva kokukhipha imodeli kusadingeka kuqinisekiswe ukuhambisana nokusebenza kwayo ngokusekelwe kudivayisi ethile nesimo sokusetshenziswa.
Ibhokisi lamathuluzi e-AIIsilinganiso 8LlamaFactory OnlineI-LlamaFactory Online iyibhokisi lamathuluzi e-AI eliku-inthanethi elenzelwe ukulungisa kahle amamodeli amakhulu avulekile, ikakhulukazi lisiza onjiniyela abazimele, amabhizinisi aqalayo namaqembu ocwaningo emanyuvesi. Abasebenzisi bangalayisha idatha ngesixhumi esibonakalayo sewebhu, bakhethe imodeli, indlela yokuqeqesha nohlelo lokunemba, bese benza ukuqeqesha, ukuqapha nemisebenzi ehlobene nokusebenzisa imodeli, ngaleyo ndlela behlise umsebenzi wokulungisa indawo ye-GPU yasendaweni. Inkundla isekela amamodeli amaningi amakhulu avulekile asetshenziswa kakhulu, ukuqeqeshwa kwangaphambilini, i-SFT, i-Reward Modeling, i-PPO, i-DPO, i-KTO nezinye izindlela zokuqeqesha, futhi inikeza izinketho zokulungisa kahle ze-LoRA, i-QLoRA nokulungisa yonke imingcele. Inkundla inezindlela zamahhala nezikhokhelwayo; ukusetshenziswa kwangempela kusadinga ukuhlola ukusetshenziswa kwamandla ekhompyutha nezindleko zokufunda kuye ngosayizi wedatha, uhlobo lwemodeli nokulungiselelwa kokuqeqesha.