[Doc] Fix doc bug and warning (#213)

* fix dgmg tutorial indentation

* bug in incidence_matrix docstring

* remove print

python/dgl/graph.py

@@ -1511,17 +1511,23 @@ class DGLGraph(object):
         value indicating whether the edge is incident to the node
         or not.
 
-        There are three types of an incidence matrix `I`:
+        There are three types of an incidence matrix :math:`I`:
+
         * "in":
-          - I[v, e] = 1 if e is the in-edge of v (or v is the dst node of e);
-          - I[v, e] = 0 otherwise.
+
+          - :math:`I[v, e] = 1` if e is the in-edge of v (or v is the dst node of e);
+          - :math:`I[v, e] = 0` otherwise.
+
         * "out":
-          - I[v, e] = 1 if e is the out-edge of v (or v is the src node of e);
-          - I[v, e] = 0 otherwise.
+
+          - :math:`I[v, e] = 1` if e is the out-edge of v (or v is the src node of e);
+          - :math:`I[v, e] = 0` otherwise.
+
         * "both":
-          - I[v, e] = 1 if e is the in-edge of v;
-          - I[v, e] = -1 if e is the out-edge of v;
-          - I[v, e] = 0 otherwise (including self-loop).
+
+          - :math:`I[v, e] = 1` if e is the in-edge of v;
+          - :math:`I[v, e] = -1` if e is the out-edge of v;
+          - :math:`I[v, e] = 0` otherwise (including self-loop).
 
         Parameters
         ----------

tutorials/models/4_rgcn.py

@@ -336,10 +336,8 @@ model.train()
 for epoch in range(n_epochs):
     optimizer.zero_grad()
     logits = model.forward(g)
-    print("after forward")
     loss = F.cross_entropy(logits[train_idx], labels[train_idx])
     loss.backward()
-    print("after backward")
 
     optimizer.step()
 

tutorials/models/5_dgmg.py

@@ -55,6 +55,7 @@ g.add_edges([2, 0], [0, 2]) # Add edges (2, 0), (0, 2)
 # with different sizes, topologies, node types, edge types, and the possibility
 # of multigraphs. Besides, a same graph can be generated in many different
 # orders. Regardless, the generative process entails a few steps:
+#
 # - Encode a changing graph,
 # - Perform actions stochastically,
 # - Collect error signals and optimize the model parameters (If we are training)