training-llms-megatron
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides technical documentation and configuration recipes for training models between 2B and 462B parameters using NVIDIA Megatron-Core.
- [SAFE]: All provided shell commands, such as those using torchrun, sbatch, and docker, are standard production practices for distributed training and model parallelism.
- [SAFE]: External dependencies such as megatron-core, torch, and transformer-engine are official, well-known libraries in the machine learning ecosystem.
- [SAFE]: All remote references point to official NVIDIA and Meta domains or repositories, which are recognized as trusted sources for this technology domain.
- [SAFE]: No evidence of prompt injection, obfuscation, unauthorized data access, or persistence mechanisms was found within the skill's instructions or reference files.
Audit Metadata