accelerate

Pass

Audited by Gen Agent Trust Hub on Aug 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill serves as a documentation reference for distributed training using HuggingFace Accelerate. All code snippets demonstrate standard usage of the library.\n- [EXTERNAL_DOWNLOADS]: The skill mentions installation of standard packages (accelerate, flash-attn) and cloning official repositories from NVIDIA (Megatron-LM, apex). These are well-known and reputable sources in the machine learning ecosystem.\n- [COMMAND_EXECUTION]: Instructions involve standard CLI tools for ML development such as pip, git, and accelerate. There are no instances of piping remote content directly into a shell or other dangerous execution patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 22, 2026, 03:31 AM
Security Audit — agent-trust-hub — accelerate