CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

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.

Quality

84%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

82%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A lean, expert-oriented reference with genuinely actionable config values, env vars, and measured numbers, plus thoughtful epistemic hygiene (rounded bands, anti-cargo-culting warnings, A/B causality caveats). The main gaps are redundant repetition of the same parallelism shapes across three sections and top-of-file references to files that are not part of the skill bundle.

Suggestions

Consolidate the TP/EP/PP/VPP shapes so they appear authoritatively in one table (e.g., the First Answer Checklist) and are referenced rather than restated in the Rounded Performance Bands and Config Families sections.

Fix the garbled run-on sentence in the First Answer Checklist ('Treat TE-scoped CUDA graph scopes ... as profile-driven candidates, CUDA_DEVICE_MAX_CONNECTIONS selection, ... and the warning not to cargo-cult tracker rows.') by recasting it as a proper list of tuning knobs.

Either include the referenced files (docs/training/moe-optimization.md, card.yaml) in the skill bundle or remove/qualify the top-of-file references so navigation does not dead-end.

DimensionReasoningScore

Conciseness

The body is dense, tabular, and assumes Claude's expertise — it explains nothing the model already knows, and the symbol legend ('E = embedding ... | = stage boundary') is only as long as needed. It sits at 4 rather than 5 because the TP/EP/PP/VPP shapes are repeated across the First Answer Checklist, Rounded Performance Bands, and Representative Config Families sections, which could be consolidated.

4 / 5

Actionability

The content is fully concrete: exact configs ('TP=2 EP=64 PP=8 VPP=4', 'Dispatcher: DeepEP'), copy-paste environment variables (CUDA_DEVICE_MAX_CONNECTIONS=1, PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True, NCCL_GRAPH_REGISTER=0), and measured results ('20.14729s/step, 299.352 model TFLOPS/GPU'). Hedges such as 'VPP=2 or similar' and the rounded bands are explicitly justified ('intentionally rounded so the document stays durable as the tracker moves'), which the rubric permits.

5 / 5

Workflow Clarity

Usage guidance is clear and ordered ('answer from these canonical rows before adding throughput caveats'; 'explicitly say VPP=unspecified; do not invent'), and a validation methodology is explicitly stated in Pitfall 7 (short screening runs, matched one-variable A/Bs, a 50-step final run). It is not a 5 because this methodology is not laid out as an explicit sequenced process with checkpoints in the main flow, and not a 3 because validation and causality separation are concretely described.

4 / 5

Progressive Disclosure

The body has good structure — clear sections, tables, and top-of-file references that are clearly signaled ('Stable docs: @docs/training/moe-optimization.md', 'Card: @skills/nemo-mbridge-perf-moe-hardware-configs/card.yaml'). However, no bundle files exist in the skill (no references/, scripts/, or assets/), the referenced paths point outside the skill, and all ~200 lines of reference tables are inlined in a single file, keeping it below anchor 5.

4 / 5

Total

17

/

20

Passed

Description

78%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, niche-specific description that states a clear 'what' with concrete usage framing and benefits from explicit trigger phrases in the frontmatter's when_to_use field. Its main weakness is that the 'when' guidance is split into a separate field rather than being self-contained in the description.

Suggestions

Fold one or two of the most natural trigger phrases from when_to_use (e.g., 'Use when asked about MoE configs or expected throughput on H100/B200/GB200-class hardware') into the description itself so it remains complete when surfaced alone.

Briefly state what the playbooks contain (parallelism/dispatcher configs and measured performance bands) to lift specificity from good to comprehensive.

DimensionReasoningScore

Specificity

The description names the domain ('Representative, point-in-time MoE training playbooks by hardware and model family') and several concrete actions ('Use them as candidate seeds, then revalidate the exact runtime, semantics, topology, and steady-state throughput'). It falls short of a 5 because it never states what the playbooks actually contain (parallelism configs, dispatcher choices).

4 / 5

Completeness

The 'what' is clear (representative point-in-time playbooks used as candidate seeds with a revalidation instruction), and an explicit 'when' exists in the frontmatter via when_to_use ('Hardware-specific MoE playbooks or throughput estimates; ...'). It is not a 5 because the trigger guidance lives in a separate field rather than the description itself — a harness surfacing only the description loses the 'when' — while the explicit guidance in the frontmatter keeps it above anchor 3.

4 / 5

Trigger Term Quality

The description carries 'MoE training playbooks', 'hardware', 'model family', and 'throughput', and the frontmatter when_to_use field adds natural phrases users would say: "'MoE on H100', 'GB200 config', 'expected throughput', 'MoE hardware playbook', 'parallelism for B200'". A few natural synonyms ('expert parallelism', 'parallelism settings') are missing, so it fits anchor 4 rather than 5.

4 / 5

Distinctiveness Conflict Risk

The description occupies a clear niche — MoE training playbooks by hardware and model family with candidate-seed revalidation — with distinct triggers like 'GB200 config' and 'MoE on H100', making conflict with other skills minimal.

5 / 5

Total

17

/

20

Passed

Validation

100%

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository

NVIDIA/skills

Path

skills/nemo-mbridge-perf-moe-hardware-configs/SKILL.md

Reviewed

Table of Contents

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.