The Agent Skills Directory

[SAFE]: The skill serves as a legitimate technical guide for model quantization. All provided code snippets follow standard machine learning practices and use well-known libraries.
[EXTERNAL_DOWNLOADS]: The skill references standard installations of Python packages such as hqq, torch, torchao, and transformers from official registries. It also points to model repositories on HuggingFace, which is expected behavior for this domain.
[COMMAND_EXECUTION]: Shell commands are restricted to standard package management (pip install) and environment verification (e.g., checking CUDA versions), presenting no security risk in this context.

hqq-quantization