GSM8K
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
GSM8K (200 samples)
66.5Exact Match Score
10
GSM8K (test)
95.3Accuracy (GSM8K)
10
GSM8K Rev
86.22Accuracy
10
GSM8K
97.66EM
10
GSM8K
80.44Accuracy
10
GSM8K
77.6Accuracy
10
GSM8K 10 (test)
97.92F1 Score
10
GSM8K
90GSM8K Score
10
GSM8K
4.818LIS
10
GSM8K
81.3Accuracy
10
GSM8K
74.66Accuracy
10
GSM8k (OOD)
72.9Accuracy
10
GSM8K 4-shot CoT
90.25Accuracy
10
GSM8K 1,000-example (test)
5.832PPL
10
GSM8k
84.4Accuracy
10
GSM8K (test)
77.3Accuracy (Base)
9
GSM8K (test)
83.55Accuracy
9
GSM8K (benign)
56ASR
9
GSM8K
85.52GSM8K Utility
9
GSM8K
70Accuracy (Clean)
9
GSM8K
8.25NFE Reduction
9
GSM8K
67.65Accuracy
9
GSM8K ID 1.0 (test)
48.8Greedy
9
GSM8K (held-out traces)
4.4FPR
9
GSM8K
2.15Accuracy Deg %
9