The Implicit Length Bias of Label Smoothing on Beam Search Decoding
원문 초록
Label smoothing is ubiquitously applied in Neural Machine Translation (NMT) training. While label smoothing offers a desired regularization effect during model training, in this paper we demonstrate that it nevertheless introduces length biases in the beam search decoding procedure. Our analysis shows that label smoothing implicitly applies a length penalty term to output sequence, causing a bias towards shorter translations. We also show that for a model fully optimized with label smoothing, translation length is implicitly upper bounded by a fixed constant independent of input. We verify our theory by applying a simple rectification function at inference time to restore the unbiased distributions from the label-smoothed model predictions. This rectification method led to consistent quality improvements on WMT English-German, English-French, English-Czech and English-Chinese tasks, up to +0.3 BLEU at beam size 4 and +2.8 BLEU at beam size 200.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Do Not Change Me: On Transferring Entities Without Modification in Neural Machine Translation -- a Multilingual Perspective
Calibrating Translation Decoding with Quality Estimation on LLMs
Beyond Decoder-only: Large Language Models Can be Good Encoders for Machine Translation
ConECT Dataset: Overcoming Data Scarcity in Context-Aware E-Commerce MT
In-Domain African Languages Translation Using LLMs and Multi-armed Bandits
THOR-MoE: Hierarchical Task-Guided and Context-Responsive Routing for Neural Machine Translation
토론
이 논문과 연결된 포럼 글이 아직 없습니다.