Shanghai AI Lab releases a new generation of scholar · Vision large model...
【数据猿导读】 Shanghai AI Lab releases a new generation of scholar · Vision large model
Recently, Shanghai Artificial Intelligence Laboratory (Shanghai AI Laboratory) jointly with Tsinghua University, the Chinese University of Hong Kong, SenseTime and other institutions open source a new generation of Scholars-vision Grand Model (InternVL). According to reports, the number of visual encoder parameters of the new generation of "Scholar · Visual basis" model reaches 6 billion (InternVL-6B), and the progressive alignment technology of contrast and generation fusion is proposed for the first time, which realizes the fine alignment of visual and language large models on Internet-level data. InternVL-6B not only processes subtle visual information in complex images and completes text generation tasks, it can also recognize and interpret information in complex pages and even solve mathematical problems.
来源:DIYuan