---

name: sar_sharpness_aware_update description: Use this skill when implementing the SAM-like sharpness-aware two-step update used by SAR recovery experiments.

Sharpness-Aware Update

When To Use

Use this skill for SAR stable test-time adaptation work that needs: Run a two-step SAM-style update around reliable entropy minimization. Do not use it as a generic ImageNet-C runner or as permission to read the original SAR repository during recovery.

Inputs

• trainable parameters, gradients or differentiable proxy model, reliable samples

Outputs

• perturbation norm, first-step parameters, second-step updated parameters, loss trace

Workflow

1. Read the caller-provided stream, logits, parameter metadata, or entropy trace.
2. Apply the SAR-specific invariant documented in this skill: SAR optimizes both entropy and entropy sharpness by minimizing entropy after an adversarial weight perturbation.
3. Produce structured numeric evidence for recovery logs.
4. Keep proxy/reduced results explicitly labeled when the full ImageNet-C runtime is unavailable.

Validation

Run python scripts/sharpness_aware_update.py when the script has a CLI, then run python -m pytest tests or the Distiller skill validator with --run-tests.

Limitations

This skill captures a reusable SAR mechanism. It does not contain pretrained weights, ImageNet-C data, or original repository code, and it should be combined with other SAR skills for end-to-end recovery.