How can I implement model?

[Byun-jinyoung]

Dear pyg users,
I am a bit new to deep learning and have studied pyg and its applications to my project.
Recently, I've read []https://arxiv.org/pdf/2106.05667.pdf which the suggested model's name is GraphiT.
I studied the paper code and wanted to implement it by using pyg. In GraphiT code, it used relative positional encoding tensor as one of the inputs generated by diffusion kernel.
But the problem what I considered is that the other inputs such as masked_x, mask with fixed length/size for each batch.
So I want to implement the model with pyg dataset/dataloader for different size tensors for each batch.

Could you please let me know how can I solve this problem?

Best regards
Thank you :)