Banking77
Banking77 is a benchmark dataset widely used for dialogue intent classification. It contains 77 intent categories related to banking operations and approximately 13,000 user utterances in total. Due to its fine-grained categories and high inter-class semantic similarity, it is regarded as a rigorous test for measuring the discriminative capability of classification models.
Timeline (last 90 days)
Both Jev and Drex support up to 255 options; Jev reaches 79.2% on the full Banking77 test (77 labels); Laya's performance degrades beyond about 20 options under default settings, reaching only 0.425 on Banking77.
Laya achieves only 0.45 accuracy on the MASSIVE dataset covering 13 languages and only 0.425 on Banking77, while JAF reports 0.87 on the same tasks.
In experiments based on the Banking77 dataset, the BGE-small + logistic regression combination achieved 93.3% accuracy, while Jev achieved only 83.2%.
Banking77 is an intent classification benchmark dataset containing 77 fine-grained banking intent categories with high inter-class semantic similarity, placing strong demands on a model's discriminative capability.
On the Banking77 intent classification dataset and multiple commercial tabular datasets, the classical models' accuracy was significantly higher than Jev's.
Banking77 is a conversational intent classification benchmark dataset containing 77 banking intent categories and a total of approximately 13,000 user utterances.
All Facts (6)
Both Jev and Drex support up to 255 options; Jev reaches 79.2% on the full Banking77 test (77 labels); Laya's performance degrades beyond about 20 options under default settings, reaching only 0.425 on Banking77.
50%UnverifiedLaya achieves only 0.45 accuracy on the MASSIVE dataset covering 13 languages and only 0.425 on Banking77, while JAF reports 0.87 on the same tasks.
50%UnverifiedIn experiments based on the Banking77 dataset, the BGE-small + logistic regression combination achieved 93.3% accuracy, while Jev achieved only 83.2%.
50%UnverifiedBanking77 is an intent classification benchmark dataset containing 77 fine-grained banking intent categories with high inter-class semantic similarity, placing strong demands on a model's discriminative capability.
50%UnverifiedOn the Banking77 intent classification dataset and multiple commercial tabular datasets, the classical models' accuracy was significantly higher than Jev's.
50%UnverifiedBanking77 is a conversational intent classification benchmark dataset containing 77 banking intent categories and a total of approximately 13,000 user utterances.
50%