Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Zero-shot Evaluation on ARC-C, ARC-E, BoolQ, HellaSwag, PIQA, RTE, and WinoGrande

72.9Accuracy

Dense

34.62844.56454.564.436Apr 6, 2026
Updated 11d ago

Evaluation Results

MethodLinks
2026.04
72.9
2026.04
70.7
2026.04
70.1
2026.04
70.1
2026.04
69.2
2026.04
68.4
2026.04
66.9
2026.04
66.2
2026.04
65.4
2026.04
65.4
2026.04
65.3
2026.04
65.3
2026.04
65
2026.04
64.3
2026.04
64
2026.04
63.9
2026.04
62.2
2026.04
61.6
2026.04
61.3
2026.04
60.8
2026.04
59
2026.04
58.9
2026.04
58.7
2026.04
58.3
2026.04
57.4
2026.04
57.1
2026.04
55.8
2026.04
55.6
2026.04
53.4
2026.04
53.3
2026.04
52.5
2026.04
52.5
2026.04
52.2
2026.04
51.7
2026.04
51.2
2026.04
49.3
2026.04
48.5
2026.04
48
2026.04
47.6
2026.04
46.9
2026.04
46.6
2026.04
46.5
2026.04
45.2
2026.04
44.3
2026.04
41.5
2026.04
41.5
2026.04
41.3
2026.04
40.8
2026.04
39.9
2026.04
39.5
2026.04
39.3
2026.04
38.5
2026.04
37.6
2026.04
37.4
2026.04
37.2
2026.04
36.6
2026.04
36.1