nemo-mbridge-recipe-recommender
Recommend and customize Megatron Bridge recipes for a user's model, GPU count, and training goal. Indexes library recipes (pretrain/SFT/PEFT) and performance recipes.
npx skills add NVIDIA/skills --skill nemo-mbridge-recipe-recommender --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
The skill helps users pick the right starting recipe and training config for Megatron Bridge by collecting user inputs (model name/size, GPU count and type, training goal, and sequence length) and then looking up the best-match recipe in its index. It then recommends the recipe function name and entry-point command, and provides adjustment guidance for parallelism resizing, batch tuning, and common pitfalls. It distinguishes library recipes (functional training) from performance recipes (throughput benchmarks), and gives a recommended first-time smoke test recipe when appropriate.
How it works
- Prompt the user for: model name/size, GPU count & type, training goal (pretrain / SFT / PEFT), and sequence length (if non-default).
- Look up the best-match recipe in the index under src/megatron/bridge/recipes/ for library recipes or the perf index under src/megatron/bridge/perf_recipes/ for perf recipes.
- Recommend the recipe function name + entry-point command.
- Provide adjustment advice: parallelism resizing, batch tuning, and potential pitfalls.
When to use it
User wants a starting recipe or training config; asks which recipe; wants to know how to train Llama; needs a starting config for X GPUs; asks what recipe for SFT.
What it can touch
Library recipes use scripts/training/run_recipe.py with --recipe <recipe_function_name> and a dataset (llm-pretrain-mock or llm-finetune). Perf recipes run via scripts/performance/run_script.py with --recipe <model_family> and --gpu_type, --num_gpus, --data mock.
Caveats
Perf recipes are for upper-bound throughput benchmarks and run on mock data by default; numbers are aspirational targets, not validated convergence configs.
# Auto Recipe — Recipe Index & Recommendation This skill indexes every shipped recipe and helps users pick the right starting config, adjust parallelism, and avoid common pitfalls. ## How to Use This Skill 1. Ask the user for: **model name/size**, **GPU count & type**, **training goal** (pretrain / SFT / PEFT), and **sequence length** (if non-default). 2. Look up the best-match recipe in the index
What does the nemo-mbridge-recipe-recommender skill do?
Recommend and customize Megatron Bridge recipes for a user's model, GPU count, and training goal. Indexes library recipes (pretrain/SFT/PEFT) and performance recipes.
How do I install it?
Run `npx skills add NVIDIA/skills --skill nemo-mbridge-recipe-recommender --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From NVIDIA/skills, a repository with 2,789 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.