The Agent Skills Directory

[SAFE]: The skill serves as a purely instructional and reference-heavy resource for machine learning practitioners. All code snippets provided are standard PyTorch/Python patterns for model training, optimization, and data processing.
[SAFE]: External references to repositories and libraries (e.g., HuggingFace, torchvision, MONAI, bitsandbytes, and Karpathy's autoresearch) are well-known and reputable sources within the machine learning ecosystem. No malicious remote script downloads or unverified dependencies were detected.
[SAFE]: The execution of commands such as nnUNetv2_train or torch.compile is consistent with the stated purpose of training neural networks and optimizing GPU performance.
[SAFE]: No obfuscation, prompt injection, or suspicious data exfiltration patterns were identified. The metadata and content are consistent with the skill's description as a technical reference guide.

ml-training-recipes