Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Question Answering on Wikidata QA
Loading...
64.5
Accuracy
Baseline (no edit)
57.74
59.495
61.25
63.005
Jun 17, 2026
Accuracy
Updated 1mo ago
Evaluation Results
Method
Method
Links
Accuracy
Baseline (no edit)
Config=–
2026.06
64.5
Redact
Config=k=10, n=6,444
2026.06
64.3
Zero ablation
Config=k=5, n=6,124
2026.06
64.3
Redact
Config=k=5, n=6,124
2026.06
64.2
Suppress
Config=k=5, n=6,124
2026.06
63
Next-best
Config=k=5, n=6,124
2026.06
62.7
Next-best
Config=k=10, n=6,444
2026.06
62
EOS
Config=k=5, n=6,124
2026.06
61.4
Suppress
Config=k=10, n=6,444
2026.06
61.3
EOS
Config=k=10, n=6,444
2026.06
59.6
Act. steering
Config=L25, α=0.5, n=6...
2026.06
58
Feedback
Search any
task
Search any
task