Эксперимент: RG Spectral Feature Gate (#579)

{ "worked": false, "confidence": 9, "verdict": "Built and numerically checked an RG spectral feature-gate MVP with correlation eigendecomposition, bandwise excess-kurtosis estimates, logarithmic slope scoring, and a synthetic downstream comparison. Noise scaling matched N^-1/2 (-0.523 observed vs -0.5 predicted), and |u4| grew approximately linearly with non-Gaussian strength, but low-exponent slope recovery was biased and the RG gate reached 0.525 accuracy versus 0.957 for top-PCA at equal retained width. Therefore the proposed mechanism and practical win were not established.", "metrics": { "baseline": "Top-16 PCA accuracy: 0.9573", "idea": "RG gate accuracy: 0.5247; noise exponent: -0.5231 observed vs -0.5 predicted; controlled slopes: [0.849, 1.389, 2.009] vs [0.5, 1.0, 2.0] predicted; |u4| by strength: [0.050, 0.160, 0.308, 0.514, 0.716]" }, "how_to_run": "/home/maxwelhelp/main/bin/python3 rg_gate_experiment.py", "files": [ "rg_gate_experiment.py", "results.json" ], "limitations": "Only a synthetic NumPy experiment was tested. No transformer, CIFAR-10, training-time straight-through gate, bootstrap/EMA calibration, FLOP or memory measurement, or minibatch mask-stability test was performed; the practical gate is a heuristic rather than the paper's full canonical-dimension flow.", "extra": "" }