Synthetic
Benchmarks
TaskDatasetSOTA resultTrendResults
Synthetic
98AUCadj
2
SYNTHETIC
62F1 macro
2
Synthetic Localized bump alternative
0.135Empirical Rejection Rate (N=300)
2
Synthetic Variance shift alternative
11.5Empirical Rejection Rate (n=300)
2
Synthetic Mean shift alternative
0.155Empirical Rejection Rate (N=300)
2
Synthetic Null setting
7.5Empirical Rejection Rate (n=300)
2
Synthetic
0.001MSE
2
Synthetic
1TOR
2
Synthetic
0.009MSE
2
Synthetic P1P2L (noiseless)
504Mean Error
2
Synthetic n=6000
30Product
2
Synthetic
7.75Communication Cost (s)
2
Synthetic
90Precision
1
Synthetic
100As-of Correctness
1
Synthetic
8Trend F1 (per-week-delta rule)
1
Synthetic
1.77Speedup
1
Synthetic
96.1mIoU
1
Synthetic (val)
0.716Rho
1
Synthetic (test)
1IR
1
Synthetic
2.7MAPE
1
Synthetic
0.48BLEU-1
1
Synthetic: Identity feature covariance and low-rank task covariance (theoretical bounds)
1n1 Parameter Value
1
Synthetic Environment Linear Scoring Noiseless σ=0 N=1000 M=1000 k=5
10Precision
1
Synthetic Category (test)
14Response Frequency
1
synthetic
—Primary metric
0