Pretrained-Language-Model
Pretrained-Language-Model copied to clipboard
wukong-为什么以ILSVRC作为验证集,模型不是在中文数据集上的训练的吗?
python main.py \ --config="configs/wukong_vit_l/wukong_vit_l.py" \ --checkpoint="/cache/ckpt/wukong_vit_l.ckpt" \ --data_dir="/cache/data/ILSVRC/"
我们同时翻译了下游任务的英文标签到中文,在主页上公开了下游数据的中文版本(https://wukong-dataset.github.io/wukong-dataset/index.html)
As mentioned in the paper, we also provide various benchmarking datasets with human-verified image-text pairs and Chinese labels for model benchmarks.