The Agent Skills Directory

[SAFE]: The skill serves as a comprehensive educational guide for machine learning architecture selection. It uses internal cross-references and documentation to assist users in choosing between different model families like CNNs, Transformers, and GNNs.
[SAFE]: All Python code snippets provided in the documentation are standard, benign implementations of neural network layers (e.g., MLP, CNN, LSTM) using the official PyTorch library.
[EXTERNAL_DOWNLOADS]: The documentation references the installation of several well-known, specialized machine learning libraries from public registries, such as flash-attn, performer-pytorch, and linformer. These are reputable third-party tools commonly used in the research community for optimizing attention mechanisms.

using-neural-architectures