The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified during the analysis of the skill. All instructions and code snippets are consistent with the stated purpose of model quantization and optimization.
[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of standard machine learning libraries (such as hqq, torch, bitblas, and torchao) and the loading of models from well-known platforms like Hugging Face. These operations target established, legitimate services.
[COMMAND_EXECUTION]: Shell commands within the skill are limited to standard Python package installation routines using pip.
[SAFE]: Employs standard framework-provided techniques for performance enhancement, such as torch.compile for kernel fusion and dynamic layer manipulation within model architectures.

hqq-quantization