domains / families / data / nemo-mbridge-recipe-recommender

nemo-mbridge-recipe-recommender

separate

Recommend and customize Megatron Bridge library and benchmark recipes for a user's model, GPU count, hardware, sequence length, and pretrain/SFT/PEFT goal. Use when selecting a starting recipe, comparing library and benchmark configs, resizing parallelism for a GPU allocation, or distinguishing convergence changes, semantics-preserving execution tuning, and benchmark-only shortcuts.

When to use

• selecting a starting recipe
• comparing library and benchmark configs
• resizing parallelism for a GPU allocation
• distinguishing convergence changes
• semantics-preserving execution tuning
• benchmark-only shortcuts
• adjusting parallelism
• avoiding common pitfalls

When not to use

• presenting benchmark recipes as production training recipes
• pairing pretraining-only mock preset with SFT or PEFT mode
• trading convergence semantics for throughput without calling it a new experiment
• using mock data losses as convergence evidence
• ranking absolute losses from different architectures directly

Skill page

Source