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

Indirect Object Identification on IOI

0.15Last-Token KL Divergence

FAP

0.09240.48120.871.2588Apr 18, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.04
0.15
2026.04
0.15
2026.04
0.22
2026.04
0.22
2026.04
0.3
2026.04
0.3
2026.04
0.52
2026.04
0.52
2026.04
0.53
2026.04
0.53
2026.04
0.64
2026.04
0.7
2026.04
0.72
2026.04
0.73
2026.04
0.74
2026.04
0.76
2026.04
0.77
2026.04
0.81
2026.04
0.82
2026.04
0.85
2026.04
0.85
2026.04
0.87
2026.04
0.91
2026.04
0.96
2026.04
1.07
2026.04
1.12
2026.04
1.13
2026.04
1.13
2026.04
1.18
2026.04
1.19
2026.04
1.22
2026.04
1.22
2026.04
1.24
2026.04
1.25
2026.04
1.26
2026.04
1.29
2026.04
1.32
2026.04
1.33
2026.04
1.52
2026.04
1.59