GitLab

* tmp fix

* add description

* rgcn with new heterograph API

* added new apply_edge()

* optimized forward pass

* renaming from *hetero to *heteroAPI
Co-authored-by: Israt Nisa <nisisrat@amazon.com>
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* Fix issue 3527

* Changed default device

* Added g to device

* Update

* Fix

* upload

* cleanup of unused code

* default gpu training/inference

* layer norm instead of batch norm

* fix for default inference mode

* simplified embedding forward method
Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>
Co-authored-by: Mufei Li <mufeili1996@gmail.com>

* first commit

* next commit

* third commit

* add ShaDow-GNN sampler and unit tests

* fixes

* lint

* cr*p

* lint

* fix lint

* fixes and more unit tests

* more tests

* fix docs

* lint

* fix

* fix

* fix

* fixes

* fix doc

* squeeze node labels in FraudDataset

* fix RLModule

* update results in README.md

* fix KeyError in full graph training
Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>

* add word_ids and simplify

* simplify

* add word_ids to be removed later

* remove word_ids

* seems to work

* tweak

* transpose word_z

* add word_ids example

* check api compatibility

* improve compatibility

* update doc

* tweak verbose

* restore word_z layout; tweak

* tweak

* tweak doc

* word_cT

* use log_weight and some other tweaks

* rewrite README

* update equations

* rewrite for clarity and pass tests

* tweak

* bugfix import

* fix unit test

* fix mult to be the same as old versions

* tweak

* could be a bugfix

* 0/0=nan

* add doc_subgraph utility function

* minor cache optimization

* minor cache tweak

* add environmental variable to trade cache speed for memory

* update README

* tweak

* add sparse update pass unit test

* simplify sparse update

* improve low-memory efficiency

* tweak

* add sample expectation scores to allow resampling

* simplify

* update comment

* avoid edge cases

* bugfix pred scores

* simplify

* add save function
Co-authored-by: Yifei Ma <yifeim@amazon.com>
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>
Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>

Remove self-loops and duplicate edges before ParMETIS and restore when converting to DGLGraph (#3472)

* save self-loops and duplicated edges separately.

* [BugFix] sort graph by dgl.ETYPE

* fix bugs in verify script

* fix verify logic

* refine README
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* [Model][Core] GATv2

* lint

* gatv2conv.py

* lint

* lint

* style and docs

* lint

* gatv2conv fix
Co-authored-by: Shaked Brody shakedbr@campus.technion.ac.il <shakedbr@tangerine.cslcs.technion.ac.il>
Co-authored-by: Mufei Li <mufeili1996@gmail.com>

* gpu compact graph template

* cuda compact graph draft

* fix typo

* compact graphs

* pass unit test but fail in training

* example using EdgeDataLoader on the GPU

* refactor cuda_compact_graph and cuda_to_block

* update training scripts

* fix linting

* fix linting

* fix exclude_edges for the GPU

* add --data-cpu & fix copyright

Co-authored-by: Jinjing Zhou <VoVAllen@users.noreply.github.com>

* Use ==/!= to compare constant literals (str, bytes, int, float, tuple)

Avoid Syntax Warnings on Python >= 3.8

$ `python3`
```
>>> "" == ""
True
>>> "" is ""
<stdin>:1: SyntaxWarning: "is" with a literal. Did you mean "=="?
True
```

* Use ==/!= to compare constant literals (str, bytes, int, float, tuple)

* The start of experiments of Jiahang Li on GraphSAINT.

* a nightly build

* a nightly build

Check the basic pipeline of codes. Next to check the details of samplers , GCN layer (forward propagation) and loss (backward propagation)

* a night build

* Implement GraphSAINT with torch.dataloader

There're still some bugs with sampling in training procedure

* Test validity

Succeed in testing validity on ppi_node experiments without testing other setup.
1. Online sampling on ppi_node experiments performs perfectly.
2. Sampling speed is a bit slow because the operations on [dgl.subgraphs], next step is to improve this part by putting the conversion into parallelism
3. Figuring out why offline+online sampling method performs bad, which does not make sense
4. Doing experiments on other setup

* Implement saint with torch.dataloader

Use torch.dataloader to speed up saint sampling with experiments. Except experiments on too large dataset Amazon, we've ...

* Added md5sum for the large dataset files

md5sum helps in validating the correctness of large dataset files once downloaded. 

Refer: https://github.com/snap-stanford/ogb/issues/253

* publish pct

* add train_cls

* add readme

* update opt for point transformer

* update the example index

* update for comments
Co-authored-by: Tong He <hetong007@gmail.com>

* Update model.py

fix typo

* Update main.py

fix autoscale

* Update README.md
Co-authored-by: Mufei Li <mufeili1996@gmail.com>

Co-authored-by: Ubuntu <ubuntu@ip-172-31-2-66.ec2.internal>

* some modifications for pointnet2

* temporarily save changes

* move files to new directory point_transformer

* implement point transformer for classification

* restore train_cls in pointnet

* implement point transformer for partseg

* fix point transformer for nan loss

* modify point transformer for cls

* modify training setting

* update transformer for cls

* update code

* update code for latest performance

* update the example index

* some minor changes
Co-authored-by: Tong He <hetong007@gmail.com>

* Update model.py

* Update README.md
Co-authored-by: Zihao Ye <expye@outlook.com>

* add evolveGCN example

* small fix

* fix defect

* fix defect
Co-authored-by: Quan (Andy) Gan <coin2028@hotmail.com>

* Create EEG-GCNN example.

* Update README.md

* Remove gitignore file.

* Update README.md

* change 'datas' to 'datasets'.

* Change train.py to main.py

* Added an entry in the indexing page.

* State "simplified version"; change how to run.

* Fix bug in contact

* Remove paper link in reference.

* Create working branch

* Add normalization of x.

* Update paper link and tags

* Update paper link in readme

* Update readme; add patient level indices

* Update readme. Add comments to models

* Update README.md

* change to with; specify location for ch and el; move note

* fix bug for note

* Add args for models; clean code.

* delete = in readme

* Add reference for spec_coh_values

* Added code for Rectifying (TypeError: unhashable type: 'slice') when copying file

* 1) added distributed preprocessing code to create ParMetis Input from CSV files
2) add code to run pm_dglpart on multiple machines
3) added support for recreating heteregenous graph from homo geneous graph based on dropped edges, as ParMetis currently only supports homogeneous graphs

* move to pandas

* Added comments and remove drop_duplicates as it was redundant

* Addressed Pr Comments

* Rename variable

* Added comment

* Added comment

* updated ReadMe
Co-authored-by: Ankit Garg <gaank@amazon.com>
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* [model] add model example GeniePath

* improvements based on feedback

* improvements based on feedback
Co-authored-by: zhjwy9343 <6593865@qq.com>

Co-authored-by: xiang song(charlie.song) <classicxsong@gmail.com>
Co-authored-by: zhjwy9343 <6593865@qq.com>

* [Model] add model example CARE-GNN

* update README

* improvements based on the review feedback

* fix missing item()
Co-authored-by: zhjwy9343 <6593865@qq.com>

* fix

* fix type
Co-authored-by: Minjie Wang <wmjlyjemaine@gmail.com>

* fix.

* fix.

* fix.

* fix.

* Fix test

* Deprecate old DistEmbedding impl, use synchronized embedding impl

* Basic imple of heterogeneous on homogenenous sampling

* make pass

* Pass C++ test

* Add python test code

* lint

* lint

* Add MultiLayerEtypeNeighborSampler

* Add unitest for single machine dataloader

* Add dist dataloader test for edge type sampler

* Fix lint

* fix

* support for per etype sample

* Fix some bug and enable distributed training with per edge sample

* fix

* Now distributed training works

* turn off some mxnet

* turn off mxnet for some dist test

* fix

* upd

* upd according to the comments

* Fix

* Fix test and now distributed works.

* upd

* upd

* Fix

* Fix bug

* remove dead code.

* upd

* Fix

* upd

* Fix
Co-authored-by: Ubuntu <ubuntu@ip-172-31-71-112.ec2.internal>
Co-authored-by: Ubuntu <ubuntu@ip-172-31-2-66.ec2.internal>
Co-authored-by: Da Zheng <zhengda1936@gmail.com>

* add model example GCN-based Anti-Spam

* update example index

* add usage info

* improvements as per comments

* fix image invisiable problem

* add image file
Co-authored-by: zhjwy9343 <6593865@qq.com>

* support hetero RGCN.

* fix.

* simplify code.

* sample_neighbors return heterograph directly.

* avoid using to_heterogeneous.

* compute canonical etypes in advance.

* fix tests.

* fix.

* fix distributed data loader for heterograph.

* use NodeDataLoader.

* fix bugs in partitioning on heterogeneous graphs.

* fix lint.

* fix tests.

* fix.

* fix.

* fix bugs.

* fix tests.

* fix.

* enable coo for distributed.

* fix.

* fix.

* fix.

* fix.

* fix.
Co-authored-by: Ubuntu <ubuntu@ip-172-31-71-112.ec2.internal>
Co-authored-by: Zheng <dzzhen@3c22fba32af5.ant.amazon.com>