* lazy eval edge ids.
* parallelize node id lookup.
* fix a bug.
* use mxnet for index.
* use update_all in the subgraph training.
* sample neighbors.
* Revert "parallelize node id lookup."
This reverts commit e661f69bb06cb5a7c246f0e106f245e27800e220.
* update README.
* cache subgraphs.
* support all degrees.
* cache adj in CPU.
* fix a bug in sse.
* print.
* raise error on mutable graphs.
* measure train time per epoch.
* fix a bug in graph_index.
* remove readonly in DGLSubGraph.
* cache subgraph properly.
* accelerate getting adjacency.
* split infer.
* lazy eval edges.
* specify inference mode.
* update for new sampler.
* use new mxnet sampling api.
* fix indent.
* remove profiling code.
* remove mxnet from sampler.
* return a lambda function for graph edges.
* add docs for immutable subgraph.
* Revert "return a lambda function for graph edges."
This reverts commit 0de5d7f100e230c518a3fb8976a6227f474d09ee.
* get parent_eid.
* rgcn tutorial
* data processing for rgcn
* many fix
* requirements for rgcn tutorial
* fix comments
* description of rgcn dataset
* author
* bug fix
* move all dataset to s3
* fix recv nodes are all 0deg; fix hybriddict does not through keyerror properly
* fallback to apply_nodes when all nodes are 0deg; WIP on pull spmv 0deg
* new 0deg behavior
* new 0deg behavior
* update mx utest for pull-0deg
* fix mx
* fix mx
* get rid of unnecessary sort-n-unique
* pickling support
* resorting to suggested way of pickling
* custom attribute pickling check
* working around a weird pytorch pickling bug
* including partial frame case
* pickling everything now
* fix as requested
* add examples in traversal.py
* message propagate methods
* use the new message propagation for tree-lstm
* update to the new name
* update propagate API doc
* update doc
* add propagate utest
* Add SH tutorials
* setup sphinx-gallery; work on graph tutorial
* draft dglgraph tutorial
* update readme to include document url
* rm obsolete file
* Draft the message passing tutorial
* Capsule code (#102)
* add capsule example
* clean code
* better naming
* better naming
* [GCN]tutorial scaffold
* fix capsule example code
* remove previous capsule example code
* graph struc edit
* modified: 2_graph.py
* update doc of capsule
* update capsule docs
* update capsule docs
* add msg passing prime
* GCN-GAT tutorial Section 1 and 2
* comment for API improvement
* section 3
* Tutorial API change (#115)
* change the API as discusses; toy example
* enable the new set/get syntax
* fixed pytorch utest
* fixed gcn example
* fixed gat example
* fixed mx utests
* fix mx utest
* delete apply edges; add utest for update_edges
* small change on toy example
* fix utest
* fix out in degrees bug
* update pagerank example and add it to CI
* add delitem for dataview
* make edges() return form that is compatible with send/update_edges etc
* fix index bug when the given data is one-int-tensor
* fix doc
* Fix
1. Fix two typos in gcn.py and gcn_spmv.py
2. Update README
* Fix GCN module
1. Update the outdated graph convolution layer class
2. Fix a bug in the code where dropout never works. Modules like dropout/batch norm depend on whether we are in the training stage or inference stage.
* Fix a bug in dropout
1. dropout depends on nn.Module.training
* Update GCN module
* Fix README
* Fix dropout & remove self.msg_field
* Fix
* Align with TF implementation
* Make g an argument for forward
* Remove features from the argument of GraphConv layer
* Support for create nodes/edges after setting representations
* Remove redundant commit
* Delete test_init_repr.py
* Test case for dynamic addition
* Base 'add_rows' upon 'append'
* Move test function
* Fix
* test by assertion
* changed add_rows to adding blank rows only; adding convert_to to backend
* moving test to basics
* oops mxnet
1. Update `examples/pytorch/gcn` and `python/dgl/nn/pytorch` based on the latest APIs
2. Add full support for dropout in `examples/pytorch/gcn` and `python/dgl/nn/pytorch`
3. Rename `GCN` class in `python/dgl/nn/pytorch` to be `GraphConvolutionLayer` class
4. Make node field an argument that can be configured by users in GraphConvolutionLayer
Note that adjacency normalization has not been supported yet in the examples.
Da Zheng