NVIDIA-NeMo
Official@nvidia-nemo
Offers a comprehensive framework for large-scale model training, inference gateway management, and modular plugin development for enterprise-grade generative systems.
Agent Skills by NVIDIA-NeMo
Showing 125 vetted skills indexed across 3 GitHub repositories.
NVIDIA-NeMonemo-mbridge-perf-moe-long-context
Guides context parallelism sizing and recompute strategies for long-context MoE training in Megatron Bridge.
NVIDIA-NeMonemo-mbridge-perf-parallelism-strategies
Guides selection and sizing of TP, PP, DP, CP, and EP parallelism configurations in Megatron Bridge.
NVIDIA-NeMonemo-mbridge-multi-node-slurm
Convert single-node training scripts into multi-node Slurm sbatch jobs and debug distributed launch failures.
NVIDIA-NeMonemo-mbridge-perf-moe-optimization-workflow
Guides evidence-gated MoE training performance optimization in Megatron Bridge.
NVIDIA-NeMonemo-mbridge-perf-tp-dp-comm-overlap
Configures TP, DP, and PP communication overlap in Megatron-Bridge training setups.
NVIDIA-NeMonemo-mbridge-resiliency
Configures fault tolerance, straggler detection, and checkpoint recovery for Megatron Bridge training.
NVIDIA-NeMonemo-mbridge-perf-moe-dispatcher-selection
Selects and validates MoE token dispatchers for Megatron Bridge training workloads.
NVIDIA-NeMonemo-mbridge-perf-hierarchical-context-parallel
Configures hierarchical context parallelism in Megatron-Bridge with a2a+p2p communication and verification steps.
NVIDIA-NeMonemo-mbridge-perf-moe-comm-overlap
Configures MoE expert-parallel communication overlap in Megatron Bridge training runs.
NVIDIA-NeMonemo-mbridge-perf-cuda-graphs
Configure and validate CUDA graph capture for Megatron Bridge training workloads.
NVIDIA-NeMonemo-mbridge-perf-expert-parallel-overlap
Validate and configure MoE expert-parallel communication overlap in Megatron-Bridge training runs.
NVIDIA-NeMonemo-mbridge-perf-activation-recompute
Configure selective and full activation recompute in Megatron Bridge to reduce GPU memory usage.
NVIDIA-NeMonemo-mbridge-memory-snapshot-analysis
Analyze and compare PyTorch CUDA memory snapshots from Megatron Bridge training runs.
NVIDIA-NeMonemo-mbridge-perf-megatron-fsdp
Configures Megatron FSDP data parallelism in Megatron-Bridge with validated settings and verification steps.
NVIDIA-NeMoverl-e2e-testing
Validates Megatron-Bridge changes through verl's Megatron backend end-to-end training runs.
NVIDIA-NeMonemo-rl-e2e-testing
Validates Megatron-Bridge model changes through external NeMo-RL end-to-end training runs.
NVIDIA-NeMoreview-pr
Reviews GitHub PRs, commits, and local diffs through staged single-agent passes.
NVIDIA-NeMonemo-mbridge-perf-sequence-packing
Configure and validate sequence packing and long-context training in Megatron-Bridge.
NVIDIA-NeMonemo-mbridge-perf-moe-hardware-configs
Provides representative MoE training configurations and throughput bands by hardware platform and model family.
NVIDIA-NeMonemo-mbridge-perf-moe-vlm-training
Guides FSDP and 3D-parallel training strategies for MoE vision-language models in Megatron Bridge.
NVIDIA-NeMonemo-mbridge-perf-nsys-analysis
Diagnose Megatron Bridge training bottlenecks from Nsight Systems traces using critical-path analysis.
NVIDIA-NeMocreate-model-verification-card
Create and validate agent-readable Megatron Bridge model verification cards in YAML.
NVIDIA-NeMonemo-mbridge-perf-memory-tuning
Diagnose and reduce peak GPU memory in Megatron Bridge training runs.
NVIDIA-NeMonemo-mbridge-recipe-recommender
Recommends and customizes Megatron Bridge training recipes for model, GPU, and training-goal combinations.
Frequently Asked Questions About NVIDIA-NeMo
FAQPage SchemaWhat specific tasks can engineers perform using the NeMo platform?▼
Engineers can execute large-scale model training, configure distributed parallelism strategies, manage inference endpoints, and build custom plugins for data processing or guardrailing within the platform.
Which personas are the primary users of these capabilities?▼
The platform targets machine learning engineers, infrastructure architects, and software developers focused on deploying, optimizing, and securing large-scale generative models within enterprise environments.
What are the core prerequisites for deploying NeMo platform services?▼
Deployment requires a GPU-accelerated environment, Docker for containerization, and familiarity with YAML-based configuration manifests for defining training pipelines, model knowledge bases, and plugin registration.