GSM8K
Benchmarks
TaskDatasetSOTA resultTrendResults
GSM8K (test)
0.06Mean Absolute Difference (|Δs|)
5
GSM8K cross-lingual
85.34RPR
5
GSM8K (cross)
99.72RPR
5
GSM8K (cross)
81.5Accuracy
5
GSM8K en
78.42Accuracy (GSM8K en)
5
GSM8K
100Accuracy under Attack (GSM8K)
5
GSM8K
92.34Verification Accuracy (VAcc)
5
GSM8K (test)
73Accuracy
5
GSM8K
5.7Accuracy Improvement (GSM8K)
5
GSM8K
90.03Accuracy (mean@8)
5
GSM8K
37.2Compression Ratio
5
GSM8K
2.04Time Per First Token (TPF)
5
GSM8k 2021 (test)
100Accuracy
5
GSM8K
98.6Exact Match (EM)
5
GSM8K
96.42ECR
5
GSM8K
91.28Accuracy
5
GSM8K
68.8Accuracy
5
GSM8K
69.9Comp. Acc.
5
GSM8K
0.964Accuracy
5
GSM8K
0.91Harmful Score (HS)
5
GSM8K
16.27Fine-tuning Accuracy (FA)
5
GSM8K
0.95Harmful Score (Clean)
5
GSM8K
5Exact Match Flexible-Extract Acc
5
GSM8K CoT 8-shot (test)
51.7TPS
5
GSM8K
93.37Accuracy
5