MGSM
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MGSM average (test)
84.8Accuracy
12
MGSM-zh (test)
89.6Accuracy
12
MGSM
88.27Accuracy
11
MGSM
6.2Accuracy
11
MGSM (test)
7.39BE
10
MGSM
87.47MGSM Score
10
MGSM Basque
85.2Accuracy
8
MGSM Code Switched P, I - (EN), Q(X) (Avg)
100Language Consistency
8
MGSM Monolingual P, I, Q - (X) (Avg)
100Language Consistency
8
MGSM Zh
69.7Pass@1
6
MGSM 18 languages
72.5Accuracy
6
MGSM
58Accuracy (Bengali)
5
MGSM
87.73Native CoT Avg Score
5
MGSM Low tier
86.2Reward
5
MGSM High tier
88.9Reward
5
MGSM Thai
87.6Score
5
MGSM-Urdu 250-problem (test)
3.87Correctness
4
MGSM Urdu
64.4Exact Match
4
MGSM (average)
60.1Accuracy (MGSM)
4
MGSM Phase-2 (test)
92Accuracy
4
MGSM Māori
41.6Accuracy
4
MGSM 10 non-English languages
76.4Non-Eng AVG Accuracy (Original)
3
MGSM Japanese
89.6Pass@1 Accuracy
2
MGSM German
91Pass@1
2
MGSM French
89.8Pass@1
2