torch-npu-pyyaml-preinstall

torch-npu fails with a ModuleNotFoundError for yaml because PyYAML is a transitive dependency not automatically installed in fresh virtual environments. Pre-installing PyYAML prevents this runtime import error in modules like memory visualization.

To prevent torch-npu transitive dependency failure, explicitly add PyYAML to your project's environment before torch-npu usage by running pip install PyYAML or updating requirements.txt to guarantee the dependency is accessible.

You need to manually install PyYAML for torch-npu when setting up a new project in a fresh virtual environment where YAML is not declared, specifically when using torch-npu memory visualization or layering dependencies via requirements.txt.

Yes, torch-npu memory visualization requires PyYAML to function correctly because it imports yaml directly. Without pre-installing PyYAML, accessing this module in a clean environment triggers a transitive dependency failure.

The best way to layer PyYAML in requirements.txt before torch-npu installation is to explicitly add PyYAML as a separate entry above torch-npu, ensuring the environment resolves the transitive dependency cleanly and prevents ModuleNotFoundError.