Deep Encoder, Shallow Decoder: Reevaluating Non-autoregressive Machine Translation · HackerPapers discussionshttps://hackerpaper.kaipfister.de/p/2006.10369/Collected HN comments and submissions linking this paper.enMon, 02 Nov 2026 11:16:22 GMTDeep Encoder, Shallow Decoder: Reevaluating Non-autoregressive Machine Translation — mentioned by srushhttps://news.ycombinator.com/item?id=38098543urn:hackerpapers:mention:2006.10369:38098543Wed, 01 Nov 2023 14:21:33 GMT<p>When distilling models for speed, you get a better win from removing decoder parameters, since they are run in serial, than encoder parameters. For example see this work https://arxiv.org/abs/2006.10369 - paper co-author</p><p><a href="https://hackerpaper.kaipfister.de/p/2006.10369/">Paper and discussion history</a>. A mention is not an endorsement.</p>cs.CL