PARD (Speculative decoding): superseded — cited as a baseline and beaten by newer methods. 2 paper(s) critique it, 1 beat it on benchmarks — #35 of 151 most-superseded. Sub-problem: cluster led by EAGLE-3. Newer alternatives in the same sub-problem include D^2SD, TreeFlash, Hybrid Verified Decoding, Bastion, Draft-OPD.

Superseded baseline#35 of 151 most-superseded

PARD

PARD: Accelerating LLM Inference with Low-Cost PARallel Draft Model Adaptation

Speculative decoding · first seen Apr 23, 2025

superseded — cited as a baseline and beaten by newer methods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 PARD as a baseline.

“However, both methods face scalability challenges when training on long sequences.”
— P-EAGLE: Parallel-Drafting EAGLE with Scalable Training
“PARD an2025pardacceleratingllminference trains small autoregressive models to mimic diffusion-style parallel generation, and then perform speculative decoding for target LLMs. However, the resulting small models lack the modeling capacity of the target LLMs, leading to limited acceptance lengths and a speedup ceiling of approximately 3×”
— DFlash: Block Diffusion for Flash Speculative Decoding

Beaten on benchmarks

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

PARD-2 beats PARD · Speedup [Q3 8B]
5.81 vs 4.39
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · tau (average acceptance length) [Q3 8B]
6.98 vs 5.56
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · Speedup [Q3 14B]
5.81 vs 4.51
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · tau (average acceptance length) [Q3 14B]
6.91 vs 5.14
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · Speedup [L3.1 8B]
5.19 vs 4.00
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · tau (average acceptance length) [L3.1 8B]
6.95 vs 5.20
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · Speedup [Q3 32B]
4.68 vs 4.37
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding
PARD-2 beats PARD · tau (average acceptance length) [Q3 32B]
5.75 vs 5.33
PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decoding

Newer alternatives

Recent methods in the same sub-problem, not yet superseded in the knowledge base.