Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Conditional Hypothesis Generation on Synthetic bill summaries Sign reversal, |H|=5
Loading...
71.6
Surface Similarity
oracle
18.976
32.638
46.3
59.962
Jun 2, 2026
Surface Similarity
F1
Updated 1mo ago
Evaluation Results
Method
Method
Links
Surface Similarity
F1
oracle
2026.06
71.6
78.8
interaction-lasso
Type=Proposed
2026.06
60
57.6
lasso
Type=SAE
2026.06
48.4
36.2
sep. score
Type=SAE
2026.06
46
37.9
demeaned-lasso (ablation)
Type=Proposed
2026.06
45.4
39.8
demeaned-reweighted-lasso
Type=Proposed
2026.06
44.8
33.5
random
2026.06
41.8
38.1
llm-covariate
Type=LLM-direct
2026.06
22.4
32.5
llm-global
Type=LLM-direct
2026.06
21
26.9
Feedback
Search any
task
Search any
task