How to get distributed training working for large dataset (> 100MM rows, ~1000 features per row)?

Setup Info
    Tags
      Help me…