GitLab

* add gin model

* convert dataset.py to data_ont_the_fly way and put it into dgl.data module

* convert dataset.py to data_ont_the_fly way and put it into dgl.data module
python code checked

* modified document and reference TUDataset; checked python part and bypass cpp part due to error

* change tensor to numpy in dataset and transform in collate@Dataloader

* Change minor format issue

Change minor format issue

* moved logging; adjusted tqdm etc

* add graph classification dataset

* add node label

* add TUDataset

* Modify to consistent with Qi Huang's implementation

* add docs

* Add docs

* Fix change of environment variable

* Update tu.py

* Update tu.py

* Fix error when add node with np.int64

* Create ppi.py

* Create train_ppi.py

* Update train_ppi.py

* Update train_ppi.py

* Create gat.py

* Update train.py

* Update train_ppi.py

* Update ppi.py

* Update train_ppi.py

* Update ppi.py

* Update train_ppi.py

* Update train_ppi.py

* Update ppi.py

* Update train_ppi.py

* update docs and readme

* batched_graph_classification

* Fix

* Fix

* update tutorial with new dataset

* Final

* Fix doc for dataset

* small fix

* minigc dataset

* more comments

* sphinx

* Fix dataset

* Track data APIs