The Devil is in the Errors: Leveraging Large Language Models for Fine-grained Machine Translation Evaluation

Keyphrases

• Devil 100%
• Large Language Models 100%
• Machine Translation Evaluation 100%
• Contextual Learning 66%
• Performance Improvement 33%
• Large Gain 33%
• Scalar 33%
• Labeled Data 33%
• Large Animal Model 33%
• Machine Translation System 33%
• Human Annotation 33%
• Quality Metrics 33%
• Iterative Development 33%
• Individual Errors 33%
• Score Prediction 33%
• Automatic Evaluation of Machine Translation 33%

Computer Science

• Machine Translation 100%
• Large Language Model 100%
• Annotation 33%
• Translation System 33%
• Interpretability 33%
• Iterative Development 33%