Gradient–Tangent Alignment Test

Fits a low-rank activation-derived tangent-subspace proxy once from pooled early-checkpoint activations, then tests whether true backpropagated gradients concentrate in it more than in a matched-rank random subspace, via a gradient-energy ratio (with a Monte Carlo null) and an IsoScore*-removal test (does ablating the proxy recover more isotropy than ablating a random subspace).