Scholar Feed

Speculative decoding

SSD

Superseded baseline#36 of 151 most-superseded

Superseded — cited as a baseline and beaten by newer methods

2 papers critique it · 1 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites SSD as a baseline.

Some approaches integrate DLLMs with speculative decoding~campbell2025self,cheng2025deerdraftdiffusionverify; however, such designs introduce autoregressive decoding behavior into DLLM inference, thereby undermining the distinctive non-autoregressive reasoning characteristics of DLLMs.
— Factorization-Error-Free Discrete Diffusion Language Model via Speculative Decoding
“SSD instead increases acceptance by adding a constant probability bias, enabling long-tail sampling but ignoring acoustic similarity and risking erroneous acceptances.”
— Principled Coarse-Grained Acceptance for Speculative Decoding in Speech

Beaten on benchmarks

Head-to-head results where a newer method reports beating SSD. Values are copied from the source paper's tables — verify against the cited paper.

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.