AI & LLM Engineering · CUDA · By NVIDIA

Shows how to launch distributed Megatron-LM training on a SLURM cluster: sbatch skeleton, torch.distributed.run setup, CUDA_DEVICE_MAX_CONNECTIONS rules and failure diagnosis.

NVIDIA/Megatron-LM

Diagnose and fix Cosmos3 environment, installation, and runtime errors.

NVIDIA/cosmos-framework

A skill your agent uses when something is wrong: Search() hangs, all evaluations return INVALIDSCORE, scores aren't improving, every config returns the same number, ptxas errors fill the log, CV% is…

NVIDIA/CompileIQ

Guide users through Cosmos3 supervised fine-tuning (SFT) post-training: preparing the example dataset and Wan2.2 VAE, converting the base checkpoint to DCP, launching distributed training (paired…

NVIDIA/cosmos-framework

A skill your agent uses when composing the Search(...) call and calling .start().

NVIDIA/CompileIQ

Turns a parquet of image file paths into a parquet of embeddings with CLIP, SigLIP or a TAO checkpoint, using the TAO Data Services container, ahead of neighbor mining.

NVIDIA/skills

Measure Jetson DRAM/NvMap usage and verify before/after memory reclamation with live audit data.

NVIDIA/skills

Pick Jetson-compatible containers, vLLM runtime images, and Jetson AI Lab PyPI indexes; maps Orin SM 8.7 vs Thor SM 11.0 and JetPack-specific package choices.

NVIDIA/skills

Continue KERMT pretraining on a custom SMILES corpus with a groverbase, cmim, or hybrid checkpoint.

NVIDIA/skills

Extract per-molecule embeddings from any encoder-bearing KERMT checkpoint.

NVIDIA/skills

Finetune a pretrained KERMT encoder on a labeled CSV. An agent skill from NVIDIA/skills.

NVIDIA/skills

A skill your agent uses when writing or debugging nvMolKit Python code for GPU-accelerated RDKit fingerprints, similarity, conformers, clustering, and molecular searches.

NVIDIA/skills

Evaluate whether an existing hot path is a credible NVIDIA Warp candidate.

NVIDIA/skills

A skill your agent uses when training, evaluating, or exporting Workflow policies with online RSL-RL or RLinf, including RL checkpoint and Workflow handoff.

NVIDIA/skills

Pretrain a fresh KERMT model from scratch on a user-provided corpus.

NVIDIA/skills

A skill your agent uses when compile time or startup time is the problem in code that uses Warp: a request to improve, optimize, or cut compile times; an app that is slow to start or stalls at the…

NVIDIA/skills

A skill your agent uses when installing, repairing, reusing, inspecting, or verifying readiness of the native NVIDIA Video Codec SDK or PyNvVideoCodec on Jetson, including the one-frame…

NVIDIA/skills

DALI imperative dynamic mode (nvidia.dali.experimental.dynamic, ndd): use when working on ndd code or migrating pipelines; skip pipeline-only tasks.

NVIDIA/skills

A skill your agent uses for CUDA-Q setup, simulation targets, QPU access, and @cudaq.kernel authoring guidance.

NVIDIA/skills

Modify, build, test, debug, and contribute to NVIDIA cuOpt (C++/CUDA, Python, server, CI).

NVIDIA/skills

NVIDIA DeepStream SDK development with Python pyservicemaker API.

NVIDIA/skills

Build deterministic forecast scripts with Earth2Studio (model, data source, IO, inference).

NVIDIA/skills

Install or repair the FoundationPose perception pipeline and build its FoundationStereo TensorRT engines.

NVIDIA/skills

Used for generating synthetic body MRI volumes with NV-Generate-CTMR rflow-mr.

NVIDIA/skills

A skill your agent uses for NVIDIA-related requests where an NVIDIA skill might help, even if the user did not ask for a skill.

NVIDIA/skills

NV-Tesseract AD Diffusion — diffusion-based anomaly detection and fine-tuning for multivariate time series.

NVIDIA/skills

NV-Tesseract Forecasting — transformer-based multivariate time series forecasting with DARR (context-enhanced kNN retrieval), interpretability, and fine-tuning.

NVIDIA/skills

A skill your agent uses when the operator is authoring, building, loading, or debugging a custom doca-bench plug-in — a versioned shared library with DOCAEXPERIMENTAL-marked C entry points that…

NVIDIA/skills

A skill your agent uses for hands-on DOCA GPI programming — wiring a GPU-Packet-Initiator context so a CUDA kernel drives RDMA queues directly from GPU memory without host CPU mediation.

NVIDIA/skills

A skill your agent uses when the user is doing hands-on DOCA GPUNetIO programming — wiring a CUDA kernel on an NVIDIA GPU to a doca-eth queue via docagpuethrxq / docagpuethtxq, standing up the…

NVIDIA/skills

A skill your agent uses when the user is building, running, or interpreting the doca/tools/gpunetioibwritebw client+server benchmark — a CUDA kernel on the client posts RDMA WRITE work requests…

NVIDIA/skills

A skill your agent uses when the user is measuring GPU-kernel-initiated RDMA WRITE latency through doca-gpunetio — building and running the gpunetioibwritelat client + server pair under…

NVIDIA/skills

Guide installing Earth2Studio via uv or pip, selecting model extras, and configuring the environment.

NVIDIA/skills

Install Holoscan SDK v4.3+ via Conda in a CUDA 13 environment.

NVIDIA/skills

Install Holoscan SDK natively on Ubuntu via apt. An agent skill from NVIDIA/skills.

NVIDIA/skills

Install Holoscan SDK Python wheel via pip into a venv. An agent skill from NVIDIA/skills.

NVIDIA/skills

Validate and use selective and full activation recompute in Megatron Bridge to reduce GPU memory usage at the cost of extra compute.

NVIDIA/skills

Used for command-shape or live NV-Reason-CXR chest X-ray reasoning smoke tests.

NVIDIA/skills

Used for running NV-Segment-CT VISTA3D on CT NIfTI volumes and recording label-map evidence.

NVIDIA/skills

Used for running NV-Segment-CTMR on CT or MRI NIfTI volumes and recording label-map evidence.

NVIDIA/skills

Create, refine, or fix NVIDIA voice agents (Cascaded or Omni) with Pipecat or LiveKit.

NVIDIA/skills

NVIDIA RAG Blueprint — deploy, configure, troubleshoot, and manage.

NVIDIA/skills

How to swap the DeepStream CV detection model in the VSS Alerts Blueprint verification (2dcv) mode - covers ONNX export, custom bbox parsers, compose mount gotchas, nvinfer config, runtime TRT…

NVIDIA/skills

Validate and use CPU offloading in Megatron Bridge, including layer-level activation offloading and fractional optimizer state offloading with HybridDeviceOptimizer.

NVIDIA/skills

Validate and use CUDA graph capture in Megatron Bridge, including local full-iteration graphs and Transformer Engine scoped graphs for attention, MLP, and MoE modules.

NVIDIA/skills

Validate and use MoE expert-parallel communication overlap in Megatron-Bridge, including overlapmoeexpertparallelcomm, delaywgradcompute, and flex dispatcher backends such as DeepEP and HybridEP.

NVIDIA/skills

Operational guide for enabling Megatron FSDP in Megatron-Bridge, including config knobs, code anchors, pitfalls, and verification.

NVIDIA/skills

Techniques for reducing peak GPU memory in Megatron Bridge — expandable segments, PEFT + SP input re-gather, parallelism resizing, activation recompute, CPU offloading constraints, and common OOM…

NVIDIA/skills