10 actions in 10 pieces of software · ranked by words, meaning and measured trust · 648 ms

1. 1nemo-mbridge-perf-moe-long-contextverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-long-context

     Long-context MoE training guidance for Megatron Bridge. Covers CP sizing, selective recompute, dispatcher choices, and practical patterns from DSV3, Qwen3, and Qwen3-Next long-context experiments.

2. 2nemo-mbridge-perf-moe-vlm-trainingverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-vlm-training

     Practical guidance for training MoE VLMs in Megatron Bridge. Compares FSDP and 3D-parallel approaches, using rounded lessons from Qwen3-VL, Qwen3-Next, and other multimodal experiments.

3. 3nemo-mbridge-perf-sequence-packingverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-sequence-packing

     Validate and use packed sequences and long-context training in Megatron-Bridge, including offline LLM packing, collate-time VLM packing, Energon online packing, and CP constraints.

4. 4nemo-mbridge-perf-moe-dispatcher-selectionverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-dispatcher-selection

     Choose the right MoE token dispatcher (`alltoall`, DeepEP, or HybridEP) for the hardware, EP degree, and optimization stage. Summarizes patterns from DSV3, Qwen3, Qwen3-Next, and VLM bring-up work.

5. 5nemo-mbridge-perf-moe-hardware-configsverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-hardware-configs

     Representative, point-in-time MoE training playbooks by hardware and model family. Use them as candidate seeds, then revalidate the exact runtime, semantics, topology, and steady-state throughput.

6. 6nemo-mbridge-perf-moe-comm-overlapverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-comm-overlap

     MoE expert-parallel communication overlap in Megatron Bridge. Covers dispatch/combine overlap, flex dispatcher backends, and expert wgrad scheduling.

7. 7nemo-mbridge-perf-expert-parallel-overlapverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-expert-parallel-overlap

     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.

8. 8nemo-mbridge-perf-cuda-graphsverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-cuda-graphs

     Validate and use CUDA graph capture in Megatron Bridge, including local full-iteration graphs and Transformer Engine scoped graphs for attention, MLP, and MoE modules.

9. 9nemo-mbridge-perf-moe-optimization-workflowverified and signedpassport

   skillnvidia/skillsrecord

   • nemo-mbridge-perf-moe-optimization-workflow

     Evidence-gated workflow for MoE performance optimization in Megatron Bridge. Covers measurement contracts, the Three Walls framework, parallel folding, profiling, matched A/B tuning, and final validation.

10. 10nemo-mbridge-perf-hierarchical-context-parallelverified and signedpassport

    skillnvidia/skillsrecord

    • nemo-mbridge-perf-hierarchical-context-parallel

      Operational guide for enabling hierarchical context parallelism in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.