SGD
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
SGD Seen domains
2.48Naturalness
4
SGD Unseen domains
2.46Naturalness
4
SGD 1.0 (overall)
28.6BLEU
4
SGD 1.0 (unseen domains)
22.2BLEU Score
4
SGD seen domains 1.0
29.4BLEU
4
SGD Synthetic (test)
81Purity
3
SGD (out-of-distribution)
61.3B-Slot Acc
3
SGD 5-shot (unseen)
2.83Grammaticality
3
SGD 5-shot (Seen)
2.83Grammaticality
3
SGD Synthetic (test)
0.826AMI
2
SGD Synthetic (test)
94.9Class-Balanced Accuracy (Full)
2
SGD in-distribution (test)
58.7B-Slot Accuracy
2
SGD in-distribution (test)
63.2B-Slot Accuracy
2
SGD
0.794Naturalness (NAT)
1
SGD
0.179Distinct-1
1