Thanks for sharing the code of your very interesting paper. I am trying to reproduce your results and was wondering if you can share the weights of the graph transformer.
Thanks for sharing the code, however I notice that in pretrain.config, the file path leads to data/all, however there is no such folder or corpus data for pre-training, which leads a significant drop in result.
Is it possible that you can update readme on pre-processing the corpus or upload pre-trained weights to guide us reproduce the result in paper? Many thanks.
Hello, I notice that you use WebQueryTable dataset, but where can I download it ?
I cannot find a download url from the original paper of WebQueryTable nor from google, although they said they released this dataset.