fix s3 link (#1310)

apps/kg/README.md

@@ -48,11 +48,11 @@ DGL-KE provides five knowledge graphs:
 
 | Dataset | #nodes | #edges | #relations |
 |---------|--------|--------|------------|
-| [FB15k](https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/FB15k.zip) | 14951 | 592213 | 1345 |
-| [FB15k-237](https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/FB15k-237.zip) | 14541 | 310116 | 237 |
-| [wn18](https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/wn18.zip) | 40943 | 151442 | 18 |
-| [wn18rr](https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/wn18rr.zip) | 40943 | 93003 | 11 |
-| [Freebase](https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/Freebase.zip) | 86054151 | 338586276 | 14824 |
+| [FB15k](https://data.dgl.ai/dataset/FB15k.zip) | 14951 | 592213 | 1345 |
+| [FB15k-237](https://data.dgl.ai/dataset/FB15k-237.zip) | 14541 | 310116 | 237 |
+| [wn18](https://data.dgl.ai/dataset/wn18.zip) | 40943 | 151442 | 18 |
+| [wn18rr](https://data.dgl.ai/dataset/wn18rr.zip) | 40943 | 93003 | 11 |
+| [Freebase](https://data.dgl.ai/dataset/Freebase.zip) | 86054151 | 338586276 | 14824 |
 
 Users can specify one of the datasets with `--dataset` in `train.py` and `eval.py`.
 

apps/kg/dataloader/KGDataset.py

@@ -38,7 +38,7 @@ class KGDataset1:
     The triples are stored as 'head_name\trelation_name\ttail_name'.
     '''
     def __init__(self, path, name, read_triple=True, only_train=False):
-        url = 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/{}.zip'.format(name)
+        url = 'https://data.dgl.ai/dataset/{}.zip'.format(name)
 
         if not os.path.exists(os.path.join(path, name)):
             print('File not found. Downloading from', url)
@@ -105,7 +105,7 @@ class KGDataset2:
     The triples are stored as 'head_nid\trelation_id\ttail_nid'.
     '''
     def __init__(self, path, name, read_triple=True, only_train=False):
-        url = 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/{}.zip'.format(name)
+        url = 'https://data.dgl.ai/dataset/{}.zip'.format(name)
 
         if not os.path.exists(os.path.join(path, name)):
             print('File not found. Downloading from', url)

docker/README.md

@@ -17,12 +17,12 @@ docker build -t dgl-lint -f Dockerfile.ci_lint .
 
 ### CPU image for kg
 ```bash
-wget https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/FB15k.zip -P install/
+wget https://data.dgl.ai/dataset/FB15k.zip -P install/
 docker build -t dgl-cpu:torch-1.2.0 -f Dockerfile.ci_cpu_torch_1.2.0 .
 ```
 
 ### GPU image for kg
 ```bash
-wget https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/FB15k.zip -P install/
+wget https://data.dgl.ai/dataset/FB15k.zip -P install/
 docker build -t dgl-gpu:torch-1.2.0 -f Dockerfile.ci_gpu_torch_1.2.0 .
 ```

examples/pytorch/graphwriter/README.md

@@ -39,6 +39,6 @@ We repeat the experiment five times.
 ### Examples
 
 We also provide the output of our implementation on test set together with the reference text.
-- [GraphWriter's output](https://s3.us-east-2.amazonaws.com/dgl.ai/models/graphwriter/tmp_pred.txt)
-- [Reference text](https://s3.us-east-2.amazonaws.com/dgl.ai/models/graphwriter/tmp_gold.txt)
+- [GraphWriter's output](https://data.dgl.ai/models/graphwriter/tmp_pred.txt)
+- [Reference text](https://data.dgl.ai/models/graphwriter/tmp_gold.txt)
 

examples/pytorch/graphwriter/prepare_data.sh

-wget https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/AGENDA.tar.gz
+wget https://data.dgl.ai/dataset/AGENDA.tar.gz
 mkdir data
 tar -C data/ -xvzf AGENDA.tar.gz

examples/pytorch/metapath2vec/download.py

@@ -24,7 +24,7 @@ class AminerDataset(object):
     """
     def __init__(self, path):
 
-        self.url = 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/aminer.zip'
+        self.url = 'https://data.dgl.ai/dataset/aminer.zip'
 
         if not os.path.exists(os.path.join(path, 'aminer.txt')):
             print('File not found. Downloading from', self.url)

examples/pytorch/model_zoo/chem/generative_models/dgmg/README.md

@@ -124,11 +124,11 @@ directory, with three statistics logged in `generation_stats.txt` under `eval_re
 
 We also provide a jupyter notebook where you can visualize the generated molecules 
 
-![](https://s3.us-east-2.amazonaws.com/dgl.ai/model_zoo/drug_discovery/dgmg/DGMG_ZINC_canonical_vis.png)
+![](https://data.dgl.ai/model_zoo/drug_discovery/dgmg/DGMG_ZINC_canonical_vis.png)
 
 and compare their property distributions against the training molecule property distributions
 
-![](https://s3.us-east-2.amazonaws.com/dgl.ai/model_zoo/drug_discovery/dgmg/DGMG_ZINC_canonical_dist.png)
+![](https://data.dgl.ai/model_zoo/drug_discovery/dgmg/DGMG_ZINC_canonical_dist.png)
 
 You can download the notebook with `wget https://data.dgl.ai/dgllife/dgmg/eval_jupyter.ipynb`.
 

examples/pytorch/model_zoo/chem/property_prediction/README.md

@@ -111,9 +111,9 @@ on the training and validation set for reference.
 
 [8] visualizes the weights of atoms in readout for possible interpretations like the figure below. 
 We provide a jupyter notebook for performing the visualization and you can download it with 
-`wget https://s3.us-east-2.amazonaws.com/dgl.ai/model_zoo/drug_discovery/AttentiveFP/atom_weight_visualization.ipynb`.
+`wget https://data.dgl.ai/model_zoo/drug_discovery/AttentiveFP/atom_weight_visualization.ipynb`.
 
-![](https://s3.us-west-2.amazonaws.com/dgl-data/dgllife/attentive_fp_vis_example.png)
+![](https://data.dgl.ai/dgllife/attentive_fp_vis_example.png)
 
 ## Dataset Customization
 

examples/pytorch/pointcloud/main.py

@@ -28,7 +28,7 @@ data_filename = 'modelnet40-sampled-2048.h5'
 local_path = args.dataset_path or os.path.join(get_download_dir(), data_filename)
 
 if not os.path.exists(local_path):
-    download('https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/modelnet40-sampled-2048.h5', local_path)
+    download('https://data.dgl.ai/dataset/modelnet40-sampled-2048.h5', local_path)
 
 CustomDataLoader = partial(
         DataLoader,

examples/pytorch/recommendation/README.md

@@ -4,7 +4,7 @@ NOTE: this version is not using NodeFlow yet.
       
 This example only work with Python 3.6+
 
-First, download and extract from https://dgl.ai.s3.us-east-2.amazonaws.com/dataset/ml-1m.tar.gz
+First, download and extract from https://data.dgl.ai/dataset/ml-1m.tar.gz
 
 One can then run the following to train PinSage on MovieLens-1M:
 

examples/pytorch/rrn/sudoku_data.py

@@ -56,7 +56,7 @@ class ListDataset(Dataset):
 
 def _get_sudoku_dataset(segment='train'):
     assert segment in ['train', 'valid', 'test']
-    url = "https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/sudoku-hard.zip"
+    url = "https://data.dgl.ai/dataset/sudoku-hard.zip"
     zip_fname = "/tmp/sudoku-hard.zip"
     dest_dir = '/tmp/sudoku-hard/'
 

examples/pytorch/rrn/sudoku_solver.py

@@ -20,7 +20,7 @@ def solve_sudoku(puzzle):
     model_filename = os.path.join(model_path, 'rrn-sudoku.pkl')
     if not os.path.exists(model_filename):
         print('Downloading model...')
-        url = 'https://s3.us-east-2.amazonaws.com/dgl.ai/models/rrn-sudoku.pkl'
+        url = 'https://data.dgl.ai/models/rrn-sudoku.pkl'
         urllib.request.urlretrieve(url, model_filename)
 
     model = torch.load(model_filename, map_location='cpu')

examples/pytorch/transformer/dataset/utils.py

@@ -4,8 +4,8 @@ import os
 from dgl.data.utils import *
 
 _urls = {
-    'wmt': 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/wmt14bpe_de_en.zip',
-    'scripts': 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/transformer_scripts.zip',
+    'wmt': 'https://data.dgl.ai/dataset/wmt14bpe_de_en.zip',
+    'scripts': 'https://data.dgl.ai/dataset/transformer_scripts.zip',
 }
 
 def prepare_dataset(dataset_name):

python/dgl/contrib/sampling/sampler.py

@@ -224,7 +224,7 @@ class NeighborSampler(NodeFlowSampler):
     layer :math:`i+1` are in layer :math:`i`. All the edges are from nodes
     in layer :math:`i` to layer :math:`i+1`.
 
-    .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/sampling/NodeFlow.png
+    .. image:: https://data.dgl.ai/tutorial/sampling/NodeFlow.png
     
     As an analogy to mini-batch training, the ``batch_size`` here is equal to the number
     of the initial seed nodes (number of nodes in the last layer).

python/dgl/nodeflow.py

@@ -82,7 +82,7 @@ class NodeFlow(DGLBaseGraph):
     We store extra information, such as the node and edge mapping from
     the NodeFlow graph to the parent graph.
 
-    .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/api/sampling.nodeflow.png
+    .. image:: https://data.dgl.ai/api/sampling.nodeflow.png
 
     DO NOT create NodeFlow object directly. Use sampling method to
     generate NodeFlow instead.

tutorials/basics/1_first.py

@@ -32,7 +32,7 @@ At the end of this tutorial, we hope you get a brief feeling of how DGL works.
 # 33). The network is visualized as follows with the color indicating the
 # community:
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/img/karate-club.png
+# .. image:: https://data.dgl.ai/tutorial/img/karate-club.png
 #    :align: center
 #
 # The task is to predict which side (0 or 33) each member tends to join given
@@ -135,7 +135,7 @@ print(G.nodes[[10, 11]].data['feat'])
 # node will update its own feature with information sent from neighboring
 # nodes. A graphical demonstration is displayed below.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/1_first/mailbox.png
+# .. image:: https://data.dgl.ai/tutorial/1_first/mailbox.png
 #    :alt: mailbox
 #    :align: center
 #
@@ -266,7 +266,7 @@ draw(0)  # draw the prediction of the first epoch
 plt.close()
 
 ###############################################################################
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/1_first/karate0.png
+# .. image:: https://data.dgl.ai/tutorial/1_first/karate0.png
 #    :height: 300px
 #    :width: 400px
 #    :align: center
@@ -278,7 +278,7 @@ plt.close()
 ani = animation.FuncAnimation(fig, draw, frames=len(all_logits), interval=200)
 
 ###############################################################################
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/1_first/karate.gif
+# .. image:: https://data.dgl.ai/tutorial/1_first/karate.gif
 #    :height: 300px
 #    :width: 400px
 #    :align: center

tutorials/basics/4_batch.py

@@ -30,7 +30,7 @@ networks to this problem has been a popular approach recently. This can be seen
 # In this tutorial, you learn how to perform batched graph classification
 # with DGL. The example task objective is to classify eight types of topologies shown here.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/dataset_overview.png
+# .. image:: https://data.dgl.ai/tutorial/batch/dataset_overview.png
 #     :align: center
 #
 # Implement a synthetic dataset :class:`data.MiniGCDataset` in DGL. The dataset has eight 
@@ -64,7 +64,7 @@ plt.show()
 # a batch of graphs can be viewed as a large graph that has many disjointed 
 # connected components. Below is a visualization that gives the general idea.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/batch.png
+# .. image:: https://data.dgl.ai/tutorial/batch/batch.png
 #     :width: 400pt
 #     :align: center
 #
@@ -91,7 +91,7 @@ def collate(samples):
 # ----------------
 # Graph classification proceeds as follows.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/graph_classifier.png
+# .. image:: https://data.dgl.ai/tutorial/batch/graph_classifier.png
 #
 # From a batch of graphs, perform message passing and graph convolution
 # for nodes to communicate with others. After message passing, compute a
@@ -254,16 +254,16 @@ print('Accuracy of argmax predictions on the test set: {:4f}%'.format(
 ###############################################################################
 # The animation here plots the probability that a trained model predicts the correct graph type.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/test_eval4.gif
+# .. image:: https://data.dgl.ai/tutorial/batch/test_eval4.gif
 #
 # To understand the node and graph representations that a trained model learned,
 # we use `t-SNE, <https://lvdmaaten.github.io/tsne/>`_ for dimensionality reduction
 # and visualization.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/tsne_node2.png
+# .. image:: https://data.dgl.ai/tutorial/batch/tsne_node2.png
 #     :align: center
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/batch/tsne_graph2.png
+# .. image:: https://data.dgl.ai/tutorial/batch/tsne_graph2.png
 #     :align: center
 #
 # The two small figures on the top separately visualize node representations after one and two

tutorials/basics/5_hetero.py

@@ -46,7 +46,7 @@ using the heterograph class and its associated API.
 # The following diagram shows several entities in the ACM dataset and the relationships among them 
 # (taken from `Shi et al., 2015 <https://arxiv.org/pdf/1511.04854.pdf>`_).
 #
-# .. figure:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/hetero/acm-example.png# 
+# .. figure:: https://data.dgl.ai/tutorial/hetero/acm-example.png# 
 # 
 # This graph has three types of entities that correspond to papers, authors, and publication venues.
 # It also contains three types of edges that connect the following:
@@ -70,7 +70,7 @@ using the heterograph class and its associated API.
 # marked with a rating, then each rating value could correspond to a different edge type.
 # The following diagram shows an example of user-item interactions as a heterograph.
 #
-# .. figure:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/hetero/recsys-example.png
+# .. figure:: https://data.dgl.ai/tutorial/hetero/recsys-example.png
 #
 #
 # Knowledge graph 
@@ -81,7 +81,7 @@ using the heterograph class and its associated API.
 # occupation (item P106) is politician (item Q82955). The relationships are shown in the following.
 # diagram.
 #
-# .. figure:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/hetero/kg-example.png
+# .. figure:: https://data.dgl.ai/tutorial/hetero/kg-example.png
 #
 
 ###############################################################################
@@ -144,7 +144,7 @@ ratings = dgl.heterograph(
 import scipy.io
 import urllib.request
 
-data_url = 'https://s3.us-east-2.amazonaws.com/dgl.ai/dataset/ACM.mat'
+data_url = 'https://data.dgl.ai/dataset/ACM.mat'
 data_file_path = '/tmp/ACM.mat'
 
 urllib.request.urlretrieve(data_url, data_file_path)

tutorials/models/1_gnn/8_sse_mx.py

@@ -567,6 +567,6 @@ for i in range(n_epochs):
 #
 # For full examples, see `Benchmark SSE on multi-GPUs <https://github.com/dmlc/dgl/tree/master/examples/mxnet/sse>`_ on Github.
 #
-# .. |image0| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/img/floodfill-paths.gif
-# .. |image1| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/img/neighbor-sampling.gif
-# .. |image2| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/img/sse.gif
+# .. |image0| image:: https://data.dgl.ai/tutorial/img/floodfill-paths.gif
+# .. |image1| image:: https://data.dgl.ai/tutorial/img/neighbor-sampling.gif
+# .. |image2| image:: https://data.dgl.ai/tutorial/img/sse.gif

tutorials/models/1_gnn/9_gat.py

@@ -55,7 +55,7 @@ structure-free normalization, in the style of attention.
 # embedding :math:`h_i^{(l+1)}` of layer :math:`l+1` from the embeddings of
 # layer :math:`l`.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/gat.png
+# .. image:: https://data.dgl.ai/tutorial/gat/gat.png
 #   :width: 450px
 #   :align: center
 #
@@ -355,7 +355,7 @@ for epoch in range(30):
 # to their labels, whereas the edges are colored according to the magnitude of
 # the attention weights, which can be referred with the colorbar on the right.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/cora-attention.png
+# .. image:: https://data.dgl.ai/tutorial/gat/cora-attention.png
 #   :width: 600px
 #   :align: center
 #
@@ -383,7 +383,7 @@ for epoch in range(30):
 #
 # As a reference, here is the histogram if all the nodes have uniform attention weight distribution.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/cora-attention-uniform-hist.png
+# .. image:: https://data.dgl.ai/tutorial/gat/cora-attention-uniform-hist.png
 #   :width: 250px
 #   :align: center
 #
@@ -453,7 +453,7 @@ for epoch in range(30):
 # learning curves of GAT and GCN are presented below; what is evident is the
 # dramatic performance adavantage of GAT over GCN.
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/ppi-curve.png
+# .. image:: https://data.dgl.ai/tutorial/gat/ppi-curve.png
 #   :width: 300px
 #   :align: center
 #
@@ -475,7 +475,7 @@ for epoch in range(30):
 #
 # Again, comparing with uniform distribution: 
 #
-# .. image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/ppi-uniform-hist.png
+# .. image:: https://data.dgl.ai/tutorial/gat/ppi-uniform-hist.png
 #   :width: 250px
 #   :align: center
 #
@@ -502,7 +502,7 @@ for epoch in range(30):
 # * See the optimized `full example <https://github.com/dmlc/dgl/blob/master/examples/pytorch/gat/gat.py>`_.
 # * The next tutorial describes how to speedup GAT models by parallelizing multiple attention heads and SPMV optimization.
 #
-# .. |image2| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/cora-attention-hist.png
-# .. |image5| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/ppi-first-layer-hist.png
-# .. |image6| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/ppi-second-layer-hist.png
-# .. |image7| image:: https://s3.us-east-2.amazonaws.com/dgl.ai/tutorial/gat/ppi-final-layer-hist.png
+# .. |image2| image:: https://data.dgl.ai/tutorial/gat/cora-attention-hist.png
+# .. |image5| image:: https://data.dgl.ai/tutorial/gat/ppi-first-layer-hist.png
+# .. |image6| image:: https://data.dgl.ai/tutorial/gat/ppi-second-layer-hist.png
+# .. |image7| image:: https://data.dgl.ai/tutorial/gat/ppi-final-layer-hist.png