GSM8K
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GSM8K
6.97Average Acceptance Length (τ)
20
GSM8K
98.5Accuracy (%)
20
GSM8K
61.33GSM8K Accuracy
20
GSM8K
90.4PRR
20
GSM8K
97.8Accuracy
20
GSM8K
78.94Exact Match Accuracy
20
GSM8k
89.7Accuracy
20
GSM8K
38.24Hit Score (HS)
20
GSM8K 1,000 examples
75.1Mean Hamming Similarity
20
GSM8K
83.85Accuracy
20
GSM8K
0FPR (%)
20
GSM8K
80.62AUROC
20
GSM8K
35.2Memorization Reduction (%)
20
GSM8K
88.16Composite Score
20
GSM8K
96Pass Rate
20
GSM8K (test)
1.619MAE (%)
20
GSM8K (test)
76.4Accuracy
20
GSM8K (full)
0.938rho
20
GSM8K
256.28Inference Speed (TPS)
20
GSM8K
0.651AUROC
20
GSM8K (test)
86.4Pass@1
20
GSM8k
85.4Accuracy
20
GSM8K
93.3Accuracy
19
GSM8K
91.74Pass@1 Accuracy
19
GSM8K
1.93Speedup (vs. AR)
19