Pronoun-Targeted Fine-tuning for NMT with Hybrid Losses

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 2267–2279, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 2267 Pronoun-Targeted Fine-tuning for NMT with Hybrid Losses Prathyusha Jwalapuram∗, Shafiq Joty∗§, Youlin Shen∗ ∗Nanyang Technological University, Singapore §Salesforce Research Asia, Singapore ∗{jwal0001,srjoty,yshen010}@ntu.edu.sg Abstract Popular Neural Machine Translation model training uses strategies like backtranslation to improve BLEU scores, requiring large amounts of additional data and training. We introduce a class of conditional generative- discriminative hybrid losses that we use to fine-tune a trained machine translation model. Through a combination of targeted fine-tuning objectives and intuitive re-use of the train- ing data the model has failed to adequately learn from, we improve the model perfor- mance of both a sentence-level and a contex- tual model without using any additional data. We target the improvement of pronoun trans- lations through our fine-tuning and evaluate our models on a pronoun benchmark testset. Our sentence-level model shows a 0.5 BLEU improvement on both the WMT14 and the IWSLT13 De-En testsets, while our contex- tual model achieves the best results, improv- ing from 31.81 to 32 BLEU on WMT14 De- En testset, and from 32.10 to 33.13 on the IWSLT13 De-En testset, with corresponding improvements in pronoun translation. We fur- ther show the generalizability of our method by reproducing the improvements on two addi- tional language pairs, Fr-En and Cs-En.1 1 Introduction The advent of neural machine translation (NMT) (Bahdanau et al., 2015; Vaswani et al., 2017) brought about significant improvements that left the previously successful statistical machine trans- lation models far behind. However, the availability of large corpora has been no small part of that suc- cess, with recent NMT models using millions of sentences for training. A lack of availability of such large parallel corpora across languages has given 1Code available at <https://github.com/ ntunlp/pronoun-finetuning>. rise to methods utilizing large amounts of mono- lingual data, such as for backtranslation (Sennrich et al., 2016a), language modeling (C¸ aglar G¨ulc¸ehre et al., 2017; Zheng et al., 2020), or for large-scale pre-training (Lewis et al., 2020). Backtranslation (Sennrich et al., 2016a; Edunov et al., 2018) is a commonly used strategy to im- prove MT models in the absence of adequate par- allel data for training. A target-to-source model is first trained using the available parallel data, which is then used to translate a large target-monolingual corpus into the source to create pseudo-parallel data for training a source-to-target MT model. This has been shown to result in improvements in the BLEU score, and has become a popular method for improving NMT models, with many recent works proposing strategies to further improve it (Hoa; Yang et al., 2019; Caswell et al., 2019). However, recent studies have suggested that there is a limit beyond which the addition of synthetic data hurts the performance of the model (Fadaee and Monz, 2018; Poncelas et al., 2018). Also, recent work (Edunov et al., 2020; Nguyen et al., 2020) point out that back-translation suffers from the trans- lationese effect, where back-translation only …

Content truncated to 500 words