HellaSwag
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HellaSwag
85.6Accuracy
35
HellaSwag 10-shot (test)
82.53Accuracy
34
HellaSwag
78.2Normalized Accuracy
33
HellaSwag
82.94Accuracy (Baseline)
31
HellaSwag 10-shot
83.3Score
27
HellaSwag 2019
80.12Normalized Accuracy
26
HellaSwag
62.67HellaSwag Zero-shot Accuracy
25
HellaSwag
3.18Expected Calibration Error (ECE)
22
M-Hellaswag
79.2Accuracy (zh)
21
HELLASWAG (test)
95.6Accuracy
21
HellaSwag (test)
0.827MAE (%)
20
HellaSwag (dev)
95.4Accuracy
20
HellaSwag
79.1Accuracy
19
HellaSwag
89.3Accuracy
19
HellaSwag 10-shot
64.85Accuracy (10-shot)
19
HellaSwag (test)
82.1Accuracy
19
Hellaswag bo
39.16Accuracy
17
HellaSwag zero-shot 200 items
68Accuracy
17
HellaSwag 1.0 (test)
85.6Accuracy
17
HellaSwag
2.38ECE
16
HellaSwag v1 (test)
81Normalized Accuracy
16
Hellaswag Multilingual (test)
83.1Accuracy
16
Hellaswag non-EU languages (test)
80.4Accuracy
16
HellaSwag
95.2Accuracy
16
HellaSwag 0-shot (N=500)
47.5Accuracy
15