The Agent Skills Directory

[SAFE]: The skill provides technical instructions and shell command templates for distributed machine learning training. No security threats were detected during the analysis of the 11 threat categories. All dependencies (megatron-core, torch, apex, transformer-engine) are standard ML libraries, and all external references point to official, well-known domains like nvidia.com and github.com. The skill's behavior strictly aligns with its stated purpose of facilitating large-scale LLM training.

training-llms-megatron