Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

CUDA graph capture in Megatron Bridge v1.0.0

Use this resource to enable and validate CUDA graph capture for Megatron Bridge training

New

  • 571 downloads
  • 249 likes
  • 247 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install cuda-graph-capture-megatron

With no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.

First time: connect the CLI with

npx --yes palatekit@latest login

The agent chooses between the local full-iteration graph and scoped Transformer Engine graphs (attention, MLP, MoE), applies the config settings and bring-up order, then checks the constraints. You get a configuration and a checklist for measuring the gain.

Mostly picked by

Updated November 2, 2026

Card for the “CUDA graph capture in Megatron Bridge” skill

Details

Generates with
Your agent's model
Licence
Apache-2.0

Similar resources

View all

What you get

  • SKILL.md: enablement method and constraints
  • card.yaml: structured skill card
  • evals/evals.json: evaluation cases

Good fit for

  • Your training is slowed by kernel launch overhead
  • You compare a full graph with per-module graphs

Not for

    Tags

    • CUDA graphs
    • performance
    • Megatron Bridge
    • training

    About

    You say

    “Enable CUDA graphs on this MoE model and tell me which mode to pick.”

    Card for the “CUDA graph capture in Megatron Bridge” skill

    Skill published by nvidia under the Apache-2.0 license, delivered in its original version.

    Show more

    How to use it

    1. Install the skill in your agent's skills folder
    2. Ask your agent to enable and validate CUDA graph capture for Megatron Bridge training
    3. Review and adjust the result

    Content preview

    Files

    • BENCHMARK.md4 KB
    • card.yaml13.8 KB
    • evals/evals.json1.4 KB
    • LICENSE11.3 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • skill-card.md4.1 KB
    • SKILL.md14.1 KB
    • skill.oms.sig4.7 KB

    Origin and licence

    Listed unmodified

    Author
    nvidia
    Source
    github.com/nvidia/skills/tree/ca116d60beafe1b4459f0ab790b7b450912c56b0/skills/nemo-mbridge-perf-cuda-graphs
    Licence
    Apache-2.0

    Notice

    « nemo-mbridge-perf-cuda-graphs » (nvidia, Apache-2.0) livré dans sa version d'origine : fiche en français et en anglais par PalateKit.

    User reviews

    –

    No ratings yet

    Loading reviews…

    Delete your review?

    Your rating and comment will be removed from the page.