A semantic-aware tokenizer that merges similar and low-entropy text spans cuts long-context token counts by up to 59% and inference latency by roughly 2x, with no reported quality loss.
Neural Machine Translation with Supervised Attention
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
The attention mechanisim is appealing for neural machine translation, since it is able to dynam- ically encode a source sentence by generating a alignment between a target word and source words. Unfortunately, it has been proved to be worse than conventional alignment models in aligment accuracy. In this paper, we analyze and explain this issue from the point view of re- ordering, and propose a supervised attention which is learned with guidance from conventional alignment models. Experiments on two Chinese-to-English translation tasks show that the super- vised attention mechanism yields better alignments leading to substantial gains over the standard attention based NMT.
fields
cs.CL 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
SemToken: Semantic-Aware Tokenization for Efficient Long-Context Language Modeling
A semantic-aware tokenizer that merges similar and low-entropy text spans cuts long-context token counts by up to 59% and inference latency by roughly 2x, with no reported quality loss.