Table 7 Updated evaluation results incorporating COMET scores for enhanced cross-lingual consistency and semantic fidelity in sentence recommendations.

From: AraTraditions10k bridging cultures with a comprehensive dataset for enhanced cross lingual image annotation retrieval and tagging

Metric

English dataset

Arabic dataset

Overall

BLEU-4

78.2

74.1

76.15

METEOR

68.3

64.6

66.45

ROUGE-L

75.8

70.4

73.1

CIDEr

136.7

133.2

134.95

SPICE

52.0

47.5

49.75

COMET

67.0

65.0

66.0

Back to article page