Эксперимент: Signed spectral attention (#1107)

{ "worked": true, "confidence": 9, "verdict": "Built signed random Fourier feature kernel and factorized linear aggregation for an analytically tractable indefinite Gaussian-mixture kernel. The mechanism manifested: signed RMSE scaled with observed slope -0.475 versus predicted -0.5, bias was near zero (0.0037 at M=512), while unsigned features converged to a nonzero RMSE floor of 0.6308. The small NumPy aggregation benchmark was slower than exact aggregation at N=220 because feature overhead dominated.", "metrics": { "baseline": "Positive-only RFF: RMSE 0.6805 at M=16 and 0.6312 at M=512, with predicted asymptotic floor 0.6308; exact aggregation median 0.00589 s.", "idea": "Signed RFF: RMSE 0.3300 at M=16 and 0.0613 at M=512, observed M^-0.475 scaling and bias 0.0037; signed factorized aggregation relative error median 0.1889 at N=220,M=128, median 0.01137 s." }, "how_to_run": "/home/maxwelhelp/main/bin/python3 signed_spectral_attention.py", "files": [ "signed_spectral_attention.py", "results.json" ], "limitations": "Only a one-dimensional analytic kernel and NumPy toy aggregation were tested; no learned transformer, CIFAR-10 accuracy, GPU implementation, positive normalization, FLOP accounting, or large-sequence scaling study was performed." }