Target
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Target n=30 (test)
65.9Macro F1
15
Target dataset
72.95Macro F1 Score
10
Target
77.3SR (%)
10
target
1ARI
7
Target qB published reference (candidate pool)
1ESSENT Score
4
target-1 citrusdrop
48Non-refusal Rate
4
Target N=262
4.09Constructive Guidance Score
2