Dataclass Migration Effort

[Phlip79]

For each dataclass, we want to:

1. Create dataclass in MLM
2. Get MLM arguments to use dataclass
3. Get dev branch in MLM to use dataclass
4. Get Bridge to use MLM dataclass

Below is a list of dataclasses we need to create:

• TrainingConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• ValidationConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• DistributedDataParallelConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• OptimizerConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• RNGConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• DistributedInitConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• RerunStateMachineConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• DataloaderConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• GPTDatasetConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• MockGPTDatasetConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• FinetuningDatasetConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• SchedulerConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• CheckpointConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• LoggerConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• ProfilingConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• TensorInspectConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• FaultToleranceConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• StragglerDetectionConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• NVRxStragglerDetectionConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass
• InProcessRestartConfig
  • 1. Create dataclass in MLM
  • 2. Get MLM arguments to use dataclass
  • 3. Get dev branch in MLM to use dataclass
  • 4. Get Bridge to use MLM dataclass

Documentation can be found here (internal-only).