Uma ufuna ukwazi ukusebenzisa izinhlelo zokusebenza zokubona komshini kukhompuyutha yakho yasekhaya, ungahle ube nentshisekelo kumodeli yolimi entsha ebizwa nge-Moodream. Iyakwazi ukucubungula okushoyo, lokho okubhalayo kanye nalokho okubonisayo. I-Moondream imodi yolimi eyinkimbinkimbi, yezinga elincane lokwenziwa kobuhlakani (AI) eletha ukusebenza okumangazayo kwemodeli encane ye-AI. Ngezilungiselelo eziyizigidi eziyizinkulungwane ezingu-1,6, i-Moodream isilungele ukuchaza kabusha indlela esisebenzisana ngayo nemishini, iyenze ibe enembile futhi iphendule izidingo zethu.
I-Moodream ayilona ithuluzi le-AI elilula; Lokhu ukugxumela phambili emkhakheni wokufunda ngomshini. Iklanyelwe ukuqonda inhlobonhlobo yedatha, okuhlanganisa ulimi olukhulunywayo, umbhalo obhaliwe nokuqukethwe okubukwayo. I-Moodream1 imodeli yolimi olubonakalayo (amapharamitha ayizigidi eziyizinkulungwane ezingu-1,6) eqeqeshwe ngu-@vikhyatk esebenza ngokulingana namamodeli aphindwe kabili ngobukhulu bayo. Iqeqeshwa kudathasethi ye-LLaVa futhi yaqalwa nge-SigLIP njengombhoshongo wokubuka kanye ne-Phi-1.5 njengesishumeki sombhalo.
Kungakhathaliseki ukuthi ungunjiniyela ofuna ukuhlanganisa i-AI kuhlelo lwakho lokusebenza, umfundi olangazelela ukufunda ubuchwepheshe bamuva, noma ungumuntu othanda i-AI, iMoodream ingeyakho. Kuyimodeli enemisebenzi eminingi ekwazi ukuguqula izinhlobo ezahlukene zolwazi zibe umbhalo noma inkulumo, ithuthukise indlela esixhumana ngayo nemishini yethu. I-Moodream imodeli yepharamitha yebhiliyoni engu-1,6 eyakhiwe kusetshenziswa i-SigLIP, Phi-1.5, kanye nedathasethi yokuqeqeshwa ye-LLaVA. Izisindo zinikezwe ilayisense ngaphansi kwe-CC-BY-SA ngenxa yokusetshenziswa kwedathasethi ye-LLaVA.
Imodeli yolimi encane ye-AI Vision 1.6B
Ukuqalisa nge-Moodream kuwumdlalo wezingane. Onjiniyela benze isiqiniseko sokuthi noma ngubani onentshisekelo kukho angayifaka kalula ngokunikeza imiyalo enemininingwane yokufaka ku-GitHub. Kungakhathaliseki ukuthi uyifaka kuphrojekthi eyinkimbinkimbi noma Uyenza ukuze ufundele wena siqu, le mihlahlandlela yenza inqubo ibe lula. Kodwa ukuzibophezela kukaMoodream kwezemfundo akugcini lapho. Ngokusebenzisana ne-Brilliant.org, ihlinzeka ngezifundo ze-AI ezisebenzisanayo ezisiza abasebenzisi baqonde futhi basebenzise amandla alobu buchwepheshe obuphambili.
Ukusebenza kwe-Moodream kuhlaba umxhwele njengokuguquguquka kwayo. Ihlolwe ngokuqinile ukuze kuqinisekiswe ukuthi ayiqondi nje kuphela idatha ngokunembile, kodwa iphendula kuyo ngokushesha. Lezi zivivinyo azifihliwe elabhorethri, zifinyeleleka kuwo wonke umuntu ku-GitHub. Lokhu kubeka izinto ngale kuvumela abasebenzisi ukuthi babe nokulindela okungokoqobo kokuthi i-Moodream ingasetshenziswa kanjani ezimweni zomhlaba wangempela, noma ngabe ukunika amandla amadivayisi ahlakaniphile asekhaya noma ukuthuthukisa ukusebenzisana kwesevisi yamakhasimende.
I-Moodream ingaphezu nje kwethuluzi; kuyisibonelo esihle senqubekelaphambili emangalisayo eyenziwa kubuchwepheshe bendawo be-AI. Lena imodeli engagcini nje ngokuphatha idatha eyinkimbinkimbi kalula, kodwa futhi inikeza imiphumela eguquguqukayo engashintshwa ukuze kusetshenziswe ububanzi obuningi. Izinsiza zemfundo ezihlinzekwa yi-Brilliant.org ziphinde zigqamise ukubaluleka kwayo, hhayi nje njengento emisha yezobuchwepheshe, kodwa futhi njengenkundla yokufunda. Ngokujoyina umphakathi nokuxhumana nabanye, ungasiza ekulolongeni ikusasa lale modeli yolimi lombono wekhompyutha. Ukuze uthole ulwazi olwengeziwe, vakashela ikhasi elisemthethweni lephrojekthi ye-GitHub.
Funda kabanzi Umhlahlandlela:
- I-BloombergGPT imodeli ye-AI yamabhiliyoni angu-50 yolimi lwezezimali
- Yimuphi usayizi ofanele wamamodeli olimi amakhulu?
- Imodeli ye-Microsoft AI itholakala ku-Azure AI - Phi-2, Orca 2, Llama 2
- Imodeli yolimi encane ye-Microsoft ye-Orca-2 13B idlula ezinye izindlela ezingama-70B
- Ungayishuna kanjani i-OpenAI's Whisper voice AI ekulotshweni
- Ungakucwengisisa kanjani ukwenza izinqumo kwe-AI ku-Semantic Router?
