[Roadmap] 0.3 Release Plan.

[classicsong]

The V0.3 release will focus on optimizing the code structure and adding new features such as multi-task learning. It will also includes other improvements and features.

• [Improvement] Graph data processing
  • [GSProcessing] Optimize spark data processing user experience
  • [GSProcessing] Support customize training/validation/test set [Feature Request] Add support of customized edge label split #761
  • [Gconstruct] Support customize training/validation/test set for edge tasks (link prediction and edge prediction) [Feature Request] Add support of customized edge label split #761
• [Improvement] Modeling and benchmarks
  • GPEFT example supports link prediction tasks
  • Add GATv2 model for homogeneous graphs
  • Add GraphStorm benchmarks using MAG and Amazon Review dataset.
• [New Feature] Multi-task support
  • [Feature Request] Support Multi-task learning #789
• [Improvement] Optimize GraphStorm code structure
  • [Change Request] Set default values to GraphStorm's gs.initialize() function #754
  • [Change Request] Redefine the role of GraphStorm's Dataset #755
  • [Change Requests] Redefine the role of GraphStorm's DataLoaders #756
  • [Change Request] Modify new Evaluator for the three tasks #757
• [Documentation] Update GraphStorm Doc
  • [Documentation] Add documentation of running gconstruct on SageMaker. #604
  • [Doc] New API example notebook 1 #830
  • [Doc] New API example Notebook 2 - Link Prediction #846
  • [Doc] Update the Notebook 0 to reflect the changes in v0.3 #844

[classicsong]

Decide to push #603 to the next release focusing on optimizing documentation.