Description: Techniques for reducing peak GPU memory in Megatron Bridge — expandable segments, PEFT + SP input re-gather, parallelism resizing, activation recompute, CPU offloading constraints, and common OOM fixes.
Lines: 327 | Code: 52 | Dir: nemo-mbridge-perf-memory-tuning
name: nemo-mbridge-perf-memory-tuning description: Techniques for reducing peak GPU memory in Megatron Bridge — expandable segments, PEFT + SP input re-gather, parallelism resizing, activation rec…