Method Drift›Speculative decoding
Superseded baseline#42 of 151 most-superseded
ParallelSpec
Speculative decoding
superseded — cited as a baseline and beaten by newer methods
3 papers critique it · 0 beat it on benchmarks
What papers say
Verbatim critique sentences, each from a paper that cites ParallelSpec as a baseline.
“Once alternatives from different depths are combined into a draft tree, they form a large combinatorial space in which many paths are not coherent continuations, and the verifier wastes budget on them.”
— SpecBlock: Block-Iterative Speculative Decoding with Dynamic Tree Drafting“these approaches still use target model information, making them inherently target-dependent”
— PARD: Accelerating LLM Inference with Low-Cost PARallel Draft Model Adaptation“ParallelSpec~xiao2024parallelspec proposed parallel drafting with a single transformer layer, but omits critical implementation details---notably whether and how target model hidden states are utilized---and does not address the memory scaling challenges that arise from extended training sequences with multiple parallel prediction positions.”
— P-EAGLE: Parallel-Drafting EAGLE with Scalable Training
Newer alternatives
Recent methods in the same sub-problem, not yet superseded in the knowledge base.
- Jun 3, 2026
- Jun 2, 2026
- Hybrid Verified DecodingHybrid Verified Decoding: Learning to Allocate Verification in Speculative DecodingMay 31, 2026
- May 28, 2026
- May 28, 2026
- May 28, 2026
- May 19, 2026
- May 19, 2026
- May 9, 2026
- May 8, 2026
- May 1, 2026
- Apr 21, 2026