CogDL v0.5.0

Release 0.5.0

The CogDL 0.5.0 release focuses on modular design and ease of use. It designs and implements a unified training loop for GNN, which introduces DataWrapper to help prepare the training/validation/test data and ModelWrapper to define the training/validation/test steps.

What's Changed

• [Bugfix] Fix MoEGCN & actnn import by @cenyk1230 in #271
• [Notebook] Add notebooks by @cenyk1230 in #276
• [Paperlist] 100 GNN papers by @cenyk1230 in #277
• [Framework] Unify the GNN training loop by @cenyk1230 in #285
• [Framework] Remove register models/datasets/wrappers by @cenyk1230 in #286
• [Pipeline] Fix pipeline by @cenyk1230 in #289
• [Custom] Fix model name by @cenyk1230 in #290
• [Docs] Update docs & examples by @cenyk1230 in #292
• [Docs] Fix building docs by @cenyk1230 in #293
• [Dataset] Update ogb arxiv & Fix epochs by @cenyk1230 in #294
• [Custom] Fix custom wrappers by @cenyk1230 in #295
• [Dataset] Add geom datasets by @cenyk1230 in #296
• [Model] Add fused GAT by @cenyk1230 in #297
• [Submodule] Add FastMoE as third-party library by @cenyk1230 in #298
• [Model] Move pyg models to examples by @cenyk1230 in #299
• [Bugfix] Fix sample adj by @cenyk1230 in #301
• [DATASET] Add description for datasets by @THINK2TRY in #304
• [Utility] Update spmm utils by @cenyk1230 in #303
• [Model] VRGCN example by @huangtinglin in #305
• [Utility] Update spmm utils by @cenyk1230 in #306
• [Bugfix] Update loading datasets by @cenyk1230 in #307
• [Feature] Support AutoGNN by @jasmine-yu in #309
• [Bugfix] Fix GAT's NaN by @cenyk1230 in #310

New Contributors

• @huangtinglin made their first contribution in #305
• @jasmine-yu made their first contribution in #309

Full Changelog: 0.4.1...v0.5.0