MoE Expert-Parallel Overlap Skill

Validate and use MoE expert-parallel communication overlap in Megatron-Bridge, including overlap_moe_expert_parallel_comm, delay_wgrad_compute, and flex dispatcher backends such as DeepEP and HybridEP.

Source: github.com/NVIDIA/skills/tree/HEAD/skills/nemo-mbridge-perf-expert-parallel-overlap · License: Other