#include <parameter_decomposition.hpp>
◆ adversarial_coefficient
| float pulsatrix::DecompositionOptions::adversarial_coefficient = 0.5f |
VPD: reconstruction under adversarial masks.
◆ divergence
◆ faithfulness_coefficient
| float pulsatrix::DecompositionOptions::faithfulness_coefficient = 1.0f |
Σ_l |W - V U|² / (number of weights): SPD's faithfulness, VPD's Δ loss (whose paper weights it 1e7).
◆ frequency_beta
| float pulsatrix::DecompositionOptions::frequency_beta = 0.5f |
VPD: the frequency term's β in Σ_c [mean_c + β mean_c log2(1 + sum_c)].
◆ importance_coefficient
| float pulsatrix::DecompositionOptions::importance_coefficient = 3e-3f |
Importance minimality's weight and exponent p (set_p() anneals it).
◆ layerwise_coefficient
| float pulsatrix::DecompositionOptions::layerwise_coefficient = 1.0f |
SPD: reconstruction with one layer masked at a time, averaged over layers.
◆ leak
| float pulsatrix::DecompositionOptions::leak = 0.01f |
The leaky slope of the causal importance's clamps.
◆ method
| float pulsatrix::DecompositionOptions::p = 1.0f |
◆ pgd_beta1
| float pulsatrix::DecompositionOptions::pgd_beta1 = 0.5f |
◆ pgd_beta2
| float pulsatrix::DecompositionOptions::pgd_beta2 = 0.99f |
◆ pgd_lr
| float pulsatrix::DecompositionOptions::pgd_lr = 0.01f |
◆ pgd_steps
| int64_t pulsatrix::DecompositionOptions::pgd_steps = 2 |
VPD's adversary: ascent steps per training step (before the one that uses the main loss's gradient), and Adam's learning rate and betas for the sources.
◆ seed
| uint64_t pulsatrix::DecompositionOptions::seed = 0 |
◆ stochastic_coefficient
| float pulsatrix::DecompositionOptions::stochastic_coefficient = 1.0f |
Reconstruction with every layer masked (SPD), or k of them (VPD; 0.5 there).
The documentation for this struct was generated from the following file: