Kronecker Regression on Synthetic Kronecker product matrices d=64
0.031LossKronMatMul
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| KronMatMuln=1024, d=64, lambda=10^-32022.09 | 0.031 | — | |
| FastKroneckerRegressionn=1024, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.0372022.09 | 0.032 | 1.051 | |
| DJSSW19n=1024, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.0372022.09 | 0.035 | 1.138 | |
| KronMatMuln=2048, d=64, lambda=10^-32022.09 | 0.123 | — | |
| FastKroneckerRegressionn=2048, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00932022.09 | 0.126 | 1.026 | |
| KronMatMuln=4096, d=64, lambda=10^-32022.09 | 0.507 | — | |
| FastKroneckerRegressionn=4096, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00232022.09 | 0.52 | 1.026 | |
| DJSSW19n=2048, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00932022.09 | 1.577 | 12.792 | |
| KronMatMuln=8192, d=64, lambda=10^-32022.09 | 2.073 | — | |
| FastKroneckerRegressionn=8192, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00062022.09 | 2.136 | 1.03 | |
| KronMatMuln=16384, d=64, lambda=10^-32022.09 | 8.238 | — | |
| FastKroneckerRegressionn=16384, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00012022.09 | 8.608 | 1.045 | |
| DJSSW19n=4096, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00232022.09 | 275.566 | 543.776 | |
| DJSSW19n=8192, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00062022.09 | 333.43 | 160.809 | |
| DJSSW19n=16384, d=64, epsilon=0.1, delta=0.01, alpha=10^-5, lambda=10^-3, Rows sampled (%)=0.00012022.09 | 546,391.728 | 66,329.791 |