OpenClaw · Skills: nvidia

NVIDIA skill for pretraining a brand-new KERMT model on a corpus you provide.

NVIDIA skill for bootstrapping the KERMT environment with Docker, GPU checks, and the image.

NVIDIA playbook for running NeMo-RL on Kubernetes, from launch through debugging.

NVIDIA skill for investigating a failed CI run and filing a GitHub issue for it.

NVIDIA skill for linting and formatting Megatron-LM code with its standard tools.

NVIDIA guide for starting distributed Megatron-LM training on SLURM.

NVIDIA skill for splitting a large PR into smaller ones to cut CODEOWNERS review groups.

NVIDIA skill for the Megatron-LM test system: layout, recipes, and unit and functional tests.

NVIDIA skill for exporting sanitized evidence from medical AI inference to MLflow.

NVIDIA skill for generating and optimizing small molecules in latent space with MolMIM.

NVIDIA skill for building protein multiple sequence alignments with ColabFold's NIM search service.

NVIDIA skill for a full protein structure prediction pipeline from MSA search to folding.

NVIDIA guide for picking and setting up distributed training strategies for AutoModel.

NVIDIA skill for configuring how NeMo AutoModel jobs launch: interactive, Slurm, or SkyPilot.

NVIDIA guide for adding new model architectures to AutoModel.

NVIDIA skill for creating and editing AutoModel recipes for training and evaluation.

NVIDIA skill for integrating NeMo Fabric into apps and platforms through its typed Python SDK.

NVIDIA skill for running Megatron-LM and Bridge training on mock or real data.

NVIDIA skill for turning single-node scripts into multi-node Slurm jobs and debugging them.

NVIDIA guide for trading compute for GPU memory with activation recompute on Megatron Bridge.