Generating several candidate translations per source sentence for knowledge distillation yields better small multilingual translators than standard single-hypothesis distillation, especially in low-resource settings.
Don't Throw Away Data: Better Sequence Knowledge Distillation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
A critical component in knowledge distillation is the means of coupling the teacher and student. The predominant sequence knowledge distillation method involves supervised learning of the student against teacher-decoded outputs, and is exemplified by the current state of the art, which incorporates minimum Bayes risk (MBR) decoding. In this paper we seek to integrate MBR more tightly in distillation training, specifically by using several high scoring MBR translations, rather than a single selected sequence, thus capturing a rich diversity of teacher outputs. Our experiments on English to German and English to Japanese translation show consistent improvements over strong baseline methods for both tasks and with varying model sizes. Additionally, we conduct a detailed analysis focusing on data efficiency and capacity curse aspects to elucidate MBR-n and explore its further potential.
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Multi-Hypothesis Distillation of Multilingual Neural Translation Models for Low-Resource Languages
Generating several candidate translations per source sentence for knowledge distillation yields better small multilingual translators than standard single-hypothesis distillation, especially in low-resource settings.