Questions about batch training with dictionary index

1

Hi, I have a quick question about how to use batch training in my data as acclecration.

My network contians input layer with different dimensions, and I use a index for each data to indicate which layer will the data be fed. However, if I intend to batch all of these data, are there any approaches I can use for batch training? If I directly use a list, the error I received is:

TypeError: unhashable type: ‘list’

Could you please help me? Thanks a lot!

2

Hi can you show model, code, training loop…

3

Hi, emm My code is like:

out = model(emblist, graphlist, graphindexlist)
all the *list represent list, and this model is a graph neural network (based on pyg).

Thanks.