Yichi Zhang
Yichi Zhang
Hi, thanks for your interest in our work. This is because the format of the raw TIVA data is complex, and we used this simple notation to label the different...
You just need to prepare the triple files and token files like existing datasets provided in `data/` and `tokens/`. The tokens are extracted by tokenizers of pre-train models like BERT...
Well, you can refer to this repo: https://github.com/microsoft/unilm/blob/master/beit2/test_get_code.py You should prepare the environments for BEiT and inference on the pre-trained codebooks of BEiT with your own images to obtain the...