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

General Capability Evaluation on ARC-C, HellaSwag, MMLU, GSM8K

54.27ARC-C Accuracy

Pre-trained

39.39843.25947.1250.981Jun 1, 2026
Updated 1d ago

Evaluation Results

MethodLinks
2026.06
54.2759.6772.6286.268.19
2026.06
53.1357.7171.283.1466.3
2026.06
52.7655.0268.9183.5465.06
2026.06
52.6557.4370.6382.8765.9
2026.06
52.5657.6170.5183.5966.07
2026.06
51.4556.8659.63751.23
2026.06
51.3955.1770.5282.7264.95
2026.06
51.2756.469.681.9764.81
2026.06
51.155.1258.2236.5850.26
2026.06
50.8555.8455.943348.91
2026.06
50.7756.4554.2631.9848.37
2026.06
50.5854.3157.3435.249.36
2026.06
50.356.1957.336.2150
2026.06
50.2457.1169.6881.2364.57
2026.06
50.1752.6353.8724.1245.2
2026.06
49.2554.2166.2477.9261.91
2026.06
49.1955.8156.7131.4448.29
2026.06
48.753.8856.3334.4448.34
2026.06
4355.8256.5427.4545.7
2026.06
42.7655.4856.2227.445.47
2026.06
42.6254.6253.5722.6143.36
2026.06
42.0254.6155.326.2244.54
2026.06
41.9453.3753.125.3843.45
2026.06
41.7752.253.942643.48
2026.06
41.6253.3250.422141.59
2026.06
40.4752.0350.6922.5841.44
2026.06
39.9752.4548.2320.5840.31