#transformer models
5 resultsToken-Level Ensemble Distillation for Grapheme-to-Phoneme Conversion
Hao Sun, Xu Tan, Jun-Wei Gan +4
The paper introduces a token‑level ensemble distillation method using Transformers to improve grapheme‑to‑phoneme conversion accuracy and reduce model size, leveraging unlabeled da…
UdS Submission for the WMT 19 Automatic Post-Editing Task
Hongfei Xu, Qiuhui Liu, Josef van Genabith
The paper presents a transformer‑based system for English‑German automatic post‑editing, adapting a context‑aware neural machine translation architecture and jointly training it wi…
Fast and Accurate Capitalization and Punctuation for Automatic Speech Recognition Using Transformer and Chunk Merging
Binh Nguyen, Vu Bao Hung Nguyen, Hien Nguyen +4
The paper introduces a transformer‑based model with a chunk‑merging strategy to jointly restore punctuation and capitalization in long automatic speech recognition transcripts, ach…
Incremental Transformer with Deliberation Decoder for Document Grounded Conversations
Zekang Li, Cheng Niu, Fandong Meng +3
The paper introduces an Incremental Transformer with a two‑pass Deliberation Decoder to generate dialogue responses that are coherent and correctly grounded in the content of a giv…
English-Czech Systems in WMT19: Document-Level Transformer
Martin Popel, Dominik MacháÄek, Michal Auersperger +2
The paper presents English‑Czech neural machine translation systems for the WMT19 news task that use a document‑level Transformer trained on overlapping multi‑sentence segments, ac…