MNLI
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
MNLI (matched)
91.7Accuracy
110
MNLI
0.32Accuracy Correlation
93
MNLI
90.8Accuracy (matched)
80
MNLI (mismatched)
91Accuracy
68
MNLI (test)
0.898Accuracy
52
MNLI (test)
91.8Clean Accuracy
44
MNLI (dev)
90.2Acc (m)
44
MNLI
88.64Accuracy
40
MNLI
86.2Accuracy
36
MNLI
2.74ECE
32
MNLI (val)
84.17Accuracy
32
MNLI mm
90.7Accuracy
30
MNLI (val)
92.13Accuracy
26
MNLI
87.4MNLI Accuracy
23
MNLI
87.98Accuracy
22
MNLI GLUE (test)
86.56Accuracy
20
MNLI
1.85ECE
16
MNLI few-shot zero-shot
71.1Accuracy
16
MNLI Neg
75Accuracy
14
MNLI
98.1Accuracy
14
MNLI
96.2Normalized Accuracy
13
MNLI-m
77.2Accuracy
13
MNLI Unknown Bias (in-distribution)
84.2Accuracy
13
MNLI HardSP (challenge)
83.2Accuracy
13
MNLI HardCD (challenge)
0.803Accuracy
13