English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Neural machine translation (NMT) models are typically trained using a softmaxcross-entropy loss where the softmax distribution is compared against smoothedgold labels. In low-resource scenarios, NMT models tend to over-fit because thesoftmax distribution quickly approaches the gold label distribution. To addressthis issue, we propose to divide the logits by a temperature coefficient, priorto applying softmax, during training. In our experiments on 11 language pairsin the Asian Language Treebank dataset and the WMT 2019 English-to-Germantranslation task, we observed significant improvements in translation qualityby up to 3.9 BLEU points. Furthermore, softmax tempering makes the greedysearch to be as good as beam search decoding in terms of translation quality,enabling 1.5 to 3.5 times speed-up. We also study the impact of softmaxtempering on multilingual NMT and recurrently stacked NMT, both of which aim toreduce the NMT model size by parameter sharing thereby verifying the utility oftemperature in developing compact NMT models. Finally, an analysis of softmaxentropies and gradients reveal the impact of our method on the internalbehavior of NMT models.

Softmax Tempering for Training Neural Machine Translation Models