Install any skill in seconds. Free to start, no credit card required.
Get Started Free →nvidia/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.
Passed 12 of 14 test cases with the skill, and 3 of 14 without it.
and 19 more in the full report
14 fixed5 passed either way3 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
I need to run the unit tests to verify the DeepEP flex dispatcher helper activation and GPU gating in the Megatron Bridge repository. What is the exact pytest command to run just the DeepEP tests quietly?
| Checked | Without skill | With skill |
|---|---|---|
| Mentions uv run python -m pytest tests/unit_tests/training/test_deepep.py -q | Missed | Passed |
The instructions your model reads when the skill loads.
Expert-parallel (EP) overlap hides the cost of token dispatch/combine all-to-all
communication by running it concurrently with expert FFN compute. Optionally,
delayed expert weight-gradient computation (delay_wgrad_compute) provides
additional overlap by deferring wgrad to overlap with the next layer's forward.
Bridge supports two dispatcher paths:
| Dispatcher | Backend | When to use |
|---|---|---|
alltoall | Standard MoE all-to-all | Default, broadest compatibility |
flex | DeepEP or HybridEP | Higher overlap on Ampere/Hopper/Blackwell |
Use EP overlap when:
EP > 1Prefer:
alltoall dispatcher for the first rollout (broader compatibility)flex + DeepEP/HybridEP when running on supported GPUs and seeking
additional gainsAvoid EP overlap when:
moe_shared_expert_overlap is enabledtest cases passed, vs 3 / 14 without
Published skills whose descriptions closely match this one, with their headline benchmark lift.
More skills from the same source and nearby in the registry, with their headline benchmark lift.