Synthetic
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Synthetic
2.88PEHE
16
Synthetic
20,864.85Prefill Throughput (tokens/s)
15
Synthetic Binary Treatment, d=200
0.194MAE
15
Synthetic Binary Treatment, d=100
0.143MAE
15
Synthetic Binary Treatment, d=50
0.052MAE
15
Synthetic Binary Treatment, d=20
0.049MAE
15
Synthetic Few-shot (Avg.)
90.7ICL Score
15
Synthetic Few-shot (P_bursty=0.1, alpha=0)
92ICL Accuracy
15
Synthetic Few-shot (P_bursty=0.9, alpha=2)
88ICL Accuracy
15
Synthetic Few-shot (P_bursty=0.9, alpha=0)
91.2ICL Accuracy
15
Synthetic Few-shot (P_bursty=0.9, alpha=1)
92.1ICL Score
15
Synthetic
0.25MSE
15
Synthetic
46.7Recovery Fraction
15
Synthetic
11.733Cumulative Regret
15
Synthetic
3.233True Rank of Reported Winner
15
Synthetic sigma=1.0
48.27Mean Accuracy
15
Synthetic sigma=0.6
58.97Mean Accuracy
15
Synthetic Euclidean model (test)
1Omega'[sat^card]
15
Synthetic MCAR, NN signal
0.144Average Computational Time (s)
14
Synthetic Hypergeometric distribution, dS=dZ=10
99.02R2
14
Synthetic Poisson distribution, dS=dZ=10
99.4R2 Score
14
Synthetic Gamma distribution, dS=dZ=10
99.35R2
14
Synthetic Negative Binomial distribution, dS=dZ=10
99.23R2 Score
14
Synthetic Gaussian distribution, dS=dZ=10
99.41R2
14
Synthetic D*
0.25PEHE
14