CaSiNo
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
CaSiNo
100IR (%)
8
CaSiNo (5-fold CV)
53.84EMA
5
CaSiNo (External)
53.84EMA
5
CaSiNo
1IR
3
CaSiNo native (held-out dialogues)
0.947Accept-F1
3
CaSiNo (test)
1IR
1
Augmented CaSiNo
60Success Rate (Simple Tasks)
1
Augmented CaSiNo
40Accuracy
1
CaSiNo
95.5Accuracy
1
CaSiNo (150-dialogue held-out split)
37.3EMA
1