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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We study lossless acceleration for seq2seq generation with a novel decodingalgorithm -- Aggressive Decoding. Unlike the previous efforts (e.g.,non-autoregressive decoding) speeding up seq2seq generation at the cost ofquality loss, our approach aims to yield the identical (or better) generationcompared with autoregressive decoding but in a significant speedup, achieved byinnovative cooperation of aggressive decoding and verification that are bothefficient due to parallel computing.  We propose two Aggressive Decoding paradigms for 2 kinds of seq2seq tasks: 1)For the seq2seq tasks whose inputs and outputs are highly similar (e.g.,Grammatical Error Correction), we propose Input-guided Aggressive Decoding(IAD) that aggressively copies from the input sentence as drafted decodedtokens to verify in parallel; 2) For other general seq2seq tasks (e.g., MachineTranslation), we propose Generalized Aggressive Decoding (GAD) that firstemploys an additional non-autoregressive decoding model for aggressive decodingand then verifies in parallel in the autoregressive manner.  We test Aggressive Decoding on the most popular 6-layer Transformer model onGPU in multiple seq2seq tasks: 1) For IAD, we show that it can introduce a7x-9x speedup for the Transformer in Grammatical Error Correction and TextSimplification tasks with the identical results as greedy decoding; 2) For GAD,we observe a 3x-5x speedup with the identical or even better quality in twoimportant seq2seq tasks: Machine Translation and Abstractive Summarization.Moreover, Aggressive Decoding can benefit even more from stronger computingdevices that are better at parallel computing. Given the lossless quality aswell as significant and promising speedup, we believe Aggressive Decoding maypotentially evolve into a de facto standard for efficient and lossless seq2seqgeneration in the near future.

Lossless Acceleration for Seq2seq Generation with Aggressive Decoding