[Doc] v0.2.1 release (#198)

README.md

@@ -97,15 +97,13 @@ from cogdl import pipeline
 stats = pipeline("dataset-stats")
 stats(["cora", "citeseer"])
 
-# visualize k-hop neighbors of seed in the dataset
-visual = pipeline("dataset-visual")
-visual("cora", seed=0, depth=3)
-
 # load OAGBert model and perform inference
 oagbert = pipeline("oagbert")
 outputs = oagbert(["CogDL is developed by KEG, Tsinghua.", "OAGBert is developed by KEG, Tsinghua."])
 ```
 
+More details of the OAGBert usage can be found [here](./cogdl/oag/README.md).
+
 ### Command-Line Usage
 
 You can also use `python scripts/train.py --task example_task --dataset example_dataset --model example_model` to run example_model on example_data and evaluate it via example_task.
@@ -144,6 +142,12 @@ Expected output:
 
 If you have ANY difficulties to get things working in the above steps, feel free to open an issue. You can expect a reply within 24 hours.
 
+### Fast-Spmm Usage
+
+CogDL provides a fast sparse matrix-matrix multiplication operator called [GE-SpMM](https://arxiv.org/abs/2007.03179) to speed up training of GNN models on the GPU. 
+You can set `fast_spmm=True` in the API usage or `--fast-spmm` in the command-line usage to enable this feature.
+Note that this feature is still in testing and may not work under some versions of CUDA.
+
 ## Docker container
 
 You might also opt to use a Docker container. There is an image available in this repo that you can build with the Torch and CUDA versions available in your system. To build the docker image just run:

README_CN.md

@@ -58,7 +58,7 @@ pip install -e .
 
 ## 使用说明
 
-### API
+### API用法
 
 您可以通过CogDL API进行各种实验，尤其是`experiment`。[quick_start.py](https://github.com/THUDM/cogdl/tree/master/examples/quick_start.py)这是一个快速入门的代码。您也可以使用自己的数据集和模型进行实验。[examples/](https://github.com/THUDM/cogdl/tree/master/examples/) 文件夹里提供了一些例子。
 
@@ -94,15 +94,12 @@ from cogdl import pipeline
 stats = pipeline("dataset-stats")
 stats(["cora", "citeseer"])
 
-# visualize k-hop neighbors of seed in the dataset
-visual = pipeline("dataset-visual")
-visual("cora", seed=0, depth=3)
-
 # load OAGBert model and perform inference
 oagbert = pipeline("oagbert")
 outputs = oagbert(["CogDL is developed by KEG, Tsinghua.", "OAGBert is developed by KEG, Tsinghua."])
 ```
 
+有关OAGBert更多的用法可以参见[这里](./cogdl/oag/README.md).
 
 ### 命令行
 基本用法可以使用 `python train.py --task example_task --dataset example_dataset --model example_model` 来在 `example_data` 上运行 `example_model` 并使用 `example_task` 来评测结果。
@@ -137,6 +134,11 @@ $ python scripts/parallel_train.py --task node_classification --dataset cora --m
 | ('cora', 'gcn') | 0.8236±0.0033 |
 | ('cora', 'gat') | 0.8262±0.0032 |
 
+### Fast-Spmm用法
+
+CogDL提供了一种快速的稀疏矩阵乘的操作（[GE-SpMM](https://arxiv.org/abs/2007.03179)）来加速图神经网络模型在GPU上的训练效率。
+你可以设置`fast_spmm=True`或者`--fast-spmm`来启用这个特性。
+需要注意的是这个特性仍在测试阶段，可能在某些CUDA版本下无法正常使用。
 
 ## Docker
 

cogdl/__init__.py

@@ -1,4 +1,4 @@
-__version__ = "0.2.0.post1"
+__version__ = "0.2.1"
 
 from .experiments import experiment
 from .oag import oagbert

cogdl/experiments.py

@@ -10,7 +10,7 @@
 
 from cogdl.options import get_default_args
 from cogdl.tasks import build_task
-from cogdl.utils import set_random_seed, tabulate_results
+from cogdl.utils import set_random_seed, tabulate_results, initialize_spmm
 from cogdl.configs import BEST_CONFIGS
 from cogdl.datasets import SUPPORTED_DATASETS
 from cogdl.models import SUPPORTED_MODELS
@@ -149,6 +149,8 @@ def raw_experiment(task: str, dataset, model, **kwargs):
     else:
         args = kwargs["args"]
 
+    initialize_spmm(args)
+
     variants = list(gen_variants(dataset=args.dataset, model=args.model, seed=args.seed))
     variants = check_task_dataset_model_match(task, variants)
 

cogdl/options.py

@@ -5,7 +5,6 @@
 from cogdl.models import MODEL_REGISTRY, try_import_model
 from cogdl.tasks import TASK_REGISTRY
 from cogdl.trainers import TRAINER_REGISTRY, try_import_trainer
-from cogdl.utils import initialize_spmm
 
 
 def get_parser():
@@ -131,7 +130,6 @@ def parse_args_and_arch(parser, args):
                 TRAINER_REGISTRY[args.trainer].add_args(parser)
     # Parse a second time.
     args = parser.parse_args()
-    initialize_spmm(args)
     return args
 
 

cogdl/utils/utils.py

@@ -223,6 +223,7 @@ def initialize_spmm(args):
 
             global fast_spmm
             fast_spmm = csrspmm
+            print("Using fast-spmm to speed up training")
         except Exception as e:
             print(e)
 

docs/source/api/datasets.rst

@@ -52,15 +52,15 @@ Matlab matrix dataset
 PyG OGB dataset
 -------------------------------
 
-.. automodule:: cogdl.datasets.pyg_ogb
+.. automodule:: cogdl.datasets.ogb
     :members:
     :undoc-members:
     :show-inheritance:
 
 PyG strategies dataset
 -------------------------------
 
-.. automodule:: cogdl.datasets.pyg_strategies_data
+.. automodule:: cogdl.datasets.strategies_data
     :members:
     :undoc-members:
     :show-inheritance:

docs/source/api/models.rst

@@ -306,11 +306,6 @@ GNN Model
     :undoc-members:
     :show-inheritance:
 
-.. autoclass:: cogdl.models.nn.fastgcn.FastGCN
-    :members:
-    :undoc-members:
-    :show-inheritance:
-
 .. autoclass:: cogdl.models.nn.mlp.MLP
     :members:
     :undoc-members:
@@ -321,7 +316,7 @@ GNN Model
     :undoc-members:
     :show-inheritance:
 
-.. autoclass:: cogdl.models.nn.pyg_stpgnn.stpgnn
+.. autoclass:: cogdl.models.nn.stpgnn.stpgnn
     :members:
     :undoc-members:
     :show-inheritance:
@@ -336,11 +331,6 @@ GNN Model
     :undoc-members:
     :show-inheritance:
 
-.. autoclass:: cogdl.models.nn.asgcn.ASGCN
-    :members:
-    :undoc-members:
-    :show-inheritance:
-
 .. autoclass:: cogdl.models.nn.dgl_gcc.GCC
     :members:
     :undoc-members:
@@ -365,7 +355,7 @@ GNN Model
 AGC Model
 ---------
 
-.. autoclass:: cogdl.models.agc.pyg_daegc.DAEGC
+.. autoclass:: cogdl.models.agc.daegc.DAEGC
     :members:
     :undoc-members:
     :show-inheritance:

docs/source/api/tasks.rst

@@ -26,14 +26,6 @@ Unsupervised Node Classification
     :undoc-members:
     :show-inheritance:
 
-Node Classification (with sampling)
--------------------------------------------
-
-.. automodule:: cogdl.tasks.node_classification_sampling
-    :members:
-    :undoc-members:
-    :show-inheritance:
-
 Heterogeneous Node Classification
 --------------------------------------------
 

docs/source/index.rst

@@ -9,19 +9,33 @@ CogDL is a graph representation learning toolkit that allows researchers and dev
 
 We summarize the contributions of CogDL as follows:
 
-  - **High Efficiency**: CogDL utilizes well-optimized operators to speed up training and save GPU memory of GNN models.
-  - **Easy-to-Use**: CogDL provides easy-to-use APIs for running experiments with the given models and datasets using hyper-parameter search.
-  - **Extensibility**: The design of CogDL makes it easy to apply GNN models to new scenarios based on our framework.
-  - **Reproducibility**: CogDL provides reproducible leaderboards for state-of-the-art models on most of important tasks in the graph domain.
-
-Supported tasks:
-
-   - Node classification
-   - Link prediction
-   - Graph classification
-   - Graph pre-training
-   - Graph clustering
-   - Graph similarity search
+- **High Efficiency**: CogDL utilizes well-optimized operators to speed up training and save GPU memory of GNN models.
+- **Easy-to-Use**: CogDL provides easy-to-use APIs for running experiments with the given models and datasets using hyper-parameter search.
+- **Extensibility**: The design of CogDL makes it easy to apply GNN models to new scenarios based on our framework.
+- **Reproducibility**: CogDL provides reproducible leaderboards for state-of-the-art models on most of important tasks in the graph domain.
+
+❗ News
+------------
+
+- We release the first version of `CogDL paper <https://arxiv.org/abs/2103.00959>`_ in arXiv. You can join `our slack <https://join.slack.com/t/cogdl/shared_invite/zt-b9b4a49j-2aMB035qZKxvjV4vqf0hEg>`_ for discussion.🎉
+- The new **v0.2.0 release** includes easy-to-use ``experiment`` and ``pipeline`` APIs for all experiments and applications. The ``experiment`` API supports automl features of searching hyper-parameters. This release also provides ``OAGBert`` API for model inference (``OAGBert`` is trained on large-scale academic corpus by our lab). Some features and models are added by the open source community (thanks to all the contributors 🎉).
+- The new **v0.1.2 release** includes a pre-training task, many examples, OGB datasets, some knowledge graph embedding methods, and some graph neural network models. The coverage of CogDL is increased to 80%. Some new APIs, such as ``Trainer`` and ``Sampler``, are developed and being tested. 
+- The new **v0.1.1 release** includes the knowledge link prediction task, many state-of-the-art models, and ``optuna`` support. We also have a `Chinese WeChat post <https://mp.weixin.qq.com/s/IUh-ctQwtSXGvdTij5eDDg>`_ about the CogDL release.
+
+Citing CogDL
+------------
+
+Please cite `our paper <https://arxiv.org/abs/2103.00959>`_ if you find our code or results useful for your research:
+
+::
+
+   @article{cen2021cogdl,
+      title={CogDL: An Extensive Toolkit for Deep Learning on Graphs},
+      author={Yukuo Cen and Zhenyu Hou and Yan Wang and Qibin Chen and Yizhen Luo and Xingcheng Yao and Aohan Zeng and Shiguang Guo and Peng Zhang and Guohao Dai and Yu Wang and Chang Zhou and Hongxia Yang and Jie Tang},
+      journal={arXiv preprint arXiv:2103.00959},
+      year={2021}
+   }
+
 
 .. toctree::
    :maxdepth: 2

docs/source/quickstart.rst

@@ -72,3 +72,11 @@ Variant                    Acc
 ('cora', 'gcn')            0.8236±0.0033  
 ('cora', 'gat')            0.8262±0.0032  
 =========================  ============== 
+
+
+Fast-Spmm Usage
+---------------
+
+CogDL provides a fast sparse matrix-matrix multiplication operator called `GE-SpMM <https://arxiv.org/abs/2007.03179>`_ to speed up training of GNN models on the GPU. 
+You can set ``fast_spmm=True`` in the API usage or ``--fast-spmm`` in the command-line usage to enable this feature.
+Note that this feature is still in testing and may not work under some versions of CUDA.