The following content is currently a work in progress and does not represent the final quality.
Alignment for the multilingual VQA tasks is being conducted on blip2-flan-t5-xxl and Guanaco using only Linear Layers.
The latest weight file is provided here, based on the implementation of MiniGPT-4.
This model supports English, Chinese, Japanese, and German languages and requires the combined use of the Guanaco 7B LLM model.
A portion of the dataset has already been released.