2026-08-06T05:13:08.006927
image/svg+xml
Matplotlib v3.11.1, https://matplotlib.org/
BLiMP
ARC-Easy
ARC-Challenge
HellaSwag
PIQA
20
30
40
50
60
70
Accuracy (%)
67.1
35.1
23.2
27.9
54.1
69.2
37.1
22.8
27.9
54.3
69.2
37.1
22.9
27.9
54.4
min-spark accuracy by effort
min-spark-low
min-spark-medium
min-spark-high