We do not need to split the dataset into batches?

chengsen / PyTorch_TextGCN

The PyTorch 1.6 and Python 3.7 implementation for the paper Graph Convolutional Networks for Text Classification

https://arxiv.org/abs/1809.05679

107 stars 21 forks source link

We do not need to split the dataset into batches? #3

Open BierOne opened 3 years ago

BierOne commented 3 years ago

Hi, chengsen.

I noticed that you directly train the model using the whole corpus, and you even do not split the corpus into batches. I am confused about this. Is this the way that the paper authors applied?

Besides, if so, how can we build the connection between the input document and the target label. I cannot find any clues if we train the model in this way.

Look forward to your reply. Thx.

regards, bill

cdqncn commented 3 years ago

GCN就是full batch方式