VideoMAE V2

VideoMAE V2 ni modeli ya video iliyoundwa na Nanjing University, Shanghai AI Lab na Shenzhen Institute of Advanced Technology (China), iliyotolewa 29 Machi 2023.

Uzito wake (weights) uko wazi: mtu yeyote anaweza kuipakua, kuiendesha na kuitumia hata kibiashara.

Ina vigezo (parameters) bilioni 1.

Ilipotoka, ilivunja rekodi katika kazi yake.

Naweza kuiendesha mwenyewe?

Ndiyo, kwenye simu janja au kifaa chenye uwezo mkubwa zaidi.

  1. SimuRAM ya GB 6–8
  2. Kompyuta mpakatoRAM ya GB 16
  3. Kompyuta yenye uwezo mkubwaGB 24 hadi 64 (kadi ya michoro au Mac yenye uwezo mkubwa)
  4. Sevazaidi ya GB 64

Inahitaji takriban GB 1.6 za kumbukumbu (RAM) ikibanwa hadi biti 4.

Programu za simu zinazoendesha modeli bila intaneti (kwa mfano PocketPal) zinapakua toleo lililobanwa.

Modeli nyingine za Nanjing University

  1. P1-235B-A22BShanghai AI LabNov 2025Lugha
    Wazi235BSeva
  2. MinerU2.5Peking University, Shanghai AI Lab na Shanghai Jiao Tong UniversitySep 2025Lugha, Utambuzi wa picha
    Wazi1.2BSimu
  3. OpenOmniChinese Academy of Sciences, National University of Singapore, Shenzhen Institute of Advanced Technology, University of Chinese Academy of Sciences na University of Science and Technology of China (USTC)Mei 2025Data mseto, Lugha
    Wazi7BKompyuta mpakato
  4. Lumina-Image-2.0Chinese University of Hong Kong (CUHK), Krea AI, Shanghai AI Lab, Shanghai Jiao Tong University na University of SydneyMar 2025Kutengeneza picha
    Wazi2.6BSimu
  5. OREAL 32BChinese University of Hong Kong (CUHK), InnoHK, Shanghai AI Lab na Shanghai Jiao Tong UniversityFeb 2025Hisabati, Lugha
    Wazi32BKompyuta yenye uwezo mkubwa
  6. OREAL 7BChinese University of Hong Kong (CUHK), InnoHK, Shanghai AI Lab na Shanghai Jiao Tong UniversityFeb 2025Hisabati, Lugha
    Wazi7BKompyuta mpakato