Add BMUF Multi-GPU Training Method

[JusperLee]

🚀 Feature

Motivation

Implements incremental block distributed data parallelism similar to https: ieeexplore.ieee.org/document/7472805. This can mitigate the performance loss caused by multi-card training.

Pitch

This can be found in the Fairseq implementation.

Additional context

---

If you enjoy Lightning, check out our other projects! ⚡

• Metrics: Machine learning metrics for distributed, scalable PyTorch applications.

• Lite: enables pure PyTorch users to scale their existing code on any kind of device while retaining full control over their own loops and optimization logic.

• Flash: The fastest way to get a Lightning baseline! A collection of tasks for fast prototyping, baselining, fine-tuning, and solving problems with deep learning.

• Bolts: Pretrained SOTA Deep Learning models, callbacks, and more for research and production with PyTorch Lightning and PyTorch.

• Lightning Transformers: Flexible interface for high-performance research using SOTA Transformers leveraging Pytorch Lightning, Transformers, and Hydra.

cc @borda @awaelchli @rohitgr7 @akihironitta

[stale[bot]]

This issue has been automatically marked as stale because it hasn't had any recent activity. This issue will be closed in 7 days if no further activity occurs. Thank you for your contributions, Pytorch Lightning Team!