SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2203.16487 · ICLR · 2022

Speculative Decoding: Exploiting Speculative Execution for Accelerating Seq2seq Generation

Tao Ge, Furu Wei, Heming Xia, Peiyi Wang, Si-Qing Chen, Zhifang Sui

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 5 functions out of this paper's own repositories and ran 5 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
hemingkx/specdec canonical 1 of 1
copy not recorded — 2 of 2
lucidrains/speculative-decoding extension 1 of 1
romsto/Speculative-Decoding reimplementation 1 of 1
FunctionStatusWhere it lives
log Ran lucidrains/speculative-decoding/speculative_decoding/speculative_decoding.py
code served (permissive licence) · get_code("b5472673332de78e")
baseline_forward_decoder Ran hemingkx/specdec/inference.py
pointer only (licence: NONE) · get_code("061c2f76d8c5307c")
default Ran this paper's copy was not recorded; identical code first harvested from ThomasMrY/VCT
pointer only · get_code("60fff7c3c400d7ff")
exists Ran this paper's copy was not recorded; identical code first harvested from ThomasMrY/VCT
pointer only · get_code("aa5486a3650902d8")
max_fn Ran romsto/Speculative-Decoding/sampling/speculative_decoding.py
code served (permissive licence) · get_code("c2abbdc78126fdbd")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

We propose Speculative Decoding (SpecDec), for the first time ever 1 , to formally study exploiting the idea of speculative execution to accelerate autoregressive (AR) decoding. Speculative Decoding has two innovations: Spec-Drafter -an independent model specially optimized for efficient and accurate drafting -and Spec-Verification -a reliable method for verifying the drafted tokens efficiently in the decoding paradigm. Experimental results on various seq2seq tasks including machine translation and abstractive summarization show our approach can achieve around 5× speedup for the popular Transformer architectures with comparable generation quality to beam search decoding, refreshing the impression that the draft-then-verify paradigm introduces only 1.4×∼2× speedup. In addition to the remarkable speedup, we also demonstrate 3 additional advantages of SpecDec, revealing its practical value for accelerating generative models in real-world applications. Our models and codes are available at https://github.com/ hemingkx/SpecDec.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2203.16487")
get_code_for_paper("2203.16487")
have("2203.16487")

Connect an agent — have() is free.