MRR: 0.7124Mean Rank: 2.96R@1: 58.28R@10: 94.45
Papers with Code · CC BY-SA 4.0
Dialogue
Dialogue is notoriously hard to evaluate. Past approaches have used human evaluation.
논문
Dialogue Success Rate: 45.8
Entity F1: 62.7
F1: 17.8BLEU-4: 1ROUGE-L: 19.3
METEOR: 0.331
MRR: 64.1Mean Rank: 4.45R@1: 50.92R@10: 88.81
Entity F1: 37.8BLEU: 12.8
Amalgamating Knowledge from Two Teachers for Task-oriented Dialogue System with Adversarial Training
Entity F1: 55.38BLEU: 17.35
MRR: 63.98Mean Rank: 4.47R@1: 50.29R@10: 88.81
Entity F1: 63.6BLEU: 17.3
MRR: 66.38Mean Rank: 4.04R@1: 53.33R@10: 90.38
Entity F1: 53.7BLEU: 13.9
MRR: 68.92Mean Rank: 3.39R@1: 55.16R@10: 92.95
Request: 97.1Joint: 88.1
Entity F1: 59.97BLEU: 14.79
Entity F1: 48.0BLEU: 13.2
Entity F1: 33.4BLEU: 12.6
Request: 96.5Joint: 84.4
Entity F1: 51.9BLEU: 12.7
BLEU-4: 20.17Score: 100.07
Request: 97.4Joint: 88.5
Entity F1: 70.07