Papers › LongSpec: Long-Context Speculative Decoding with Efficient Drafting and Verification

LongSpec: Long-Context Speculative Decoding with Efficient Drafting and Verification

24 Feb 2025arXiv:2502.17421archive 2025-07-28

Penghui Yang, Cunxiao Du, Fengzhuo Zhang, Haonan Wang, Tianyu Pang, Chao Du, Bo An

Speculative decoding has become a promising technique to mitigate the high inference latency of autoregressive decoding in Large Language Models (LLMs). Despite its promise, the effective application of speculative decoding in LLMs still confronts three key challenges: the increasing memory demands of the draft model, the distribution shift between the short-training corpora and long-context inference, and inefficiencies in attention implementation. In this work, we enhance the performance of speculative decoding in long-context settings by addressing these challenges. First, we propose a memory-efficient draft model with a constant-sized Key-Value (KV) cache. Second, we introduce novel position indices for short-training data, enabling seamless adaptation from short-context training to long-context inference. Finally, we present an innovative attention aggregation method that combines fast implementations for prefix computation with standard attention for tree mask handling, effectively resolving the latency and memory inefficiencies of tree decoding. Our approach achieves strong results on various long-context tasks, including repository-level code completion, long-context summarization, and o1-like long reasoning tasks, demonstrating significant improvements in latency reduction. The code is available at https://github.com/sail-sg/LongSpec.

PaperPDFCodeCode Syntology ran

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

For agents, Syntology's MCP tool lists every function and class Syntology harvested from this paper and whether it ran (how to connect): get_harvested_code_for_paper(arxiv_id="2502.17421")

Code

Syntology Ran 2 of 10 code samples harvested from 1 repository linked to this paper; 8 have no recorded run. Of those that ran: 2 ran · our draft was wrong.

By repository: official repository: 10 samples from 1 repository, 2 ran. The run record, sample by sample. “Ran” means executed on a synthesized input, not that the code is correct or reproduces the paper.

sail-sg/longspec officialmentioned in papermentioned on GitHubpytorchMIT report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

10 samples harvested; 2 ran; 0 honoured the contract we drafted; 8 have no recorded run. Read from Syntology's graph 2026-09-24; that is when this build read the record, not when the samples ran.

2ran · our draft was wrong
8unverified

Licence: 0 of the 10 samples are pointer only, meaning Syntology does not serve that copy's text. This page shows no code text for any sample; each one links to its file in the repository.

Harvested from sail-sg/longspec. “Ran” means the sample executed on a synthesized input. It does not mean the output is correct, and nothing here reproduces the paper's results. “Honoured” and “violated” refer to a contract Syntology drafted from the code itself; “our draft was wrong” and “fixture could not drive it” are failures of Syntology's instrument, not of the code.

Each sample ends with its code_sha256, Syntology's identity for that exact code. An agent fetches the stored sample with Syntology's MCP tool get_code(code_sha256="…") (how to connect); click an identity to copy that call.

Repository labels, per sample. official repository: The archive marks this repository official for the paper. named in the paper: The archive records that the paper mentions this repository; it is not marked official. community (archive-listed): In the archive's code links for this paper, not marked official and not recorded as mentioned in the paper. found in paper text by Syntology: Syntology found this repository in the paper's own text; whether it is the authors' implementation is not asserted. community: Not in the archive's code links for this paper; a community repository Syntology harvested. Samples from a repository marked official are listed first. Licence labels name the repository's licence as recorded at harvest. “Pointer only” means Syntology does not serve that copy's text, for one of four reasons: no licence file was found; the licence was not identified; the licence is recorded as permissive but that copy's record is not marked cleared; or the licence is outside the permissive list Syntology serves text under (MIT, Apache-2.0, BSD and similar). Some licences outside that list permit redistribution, such as WTFPL, and GPL-3.0 under its conditions; they are simply not on the list. Hover a licence label for the reason. File links open the file on GitHub at the default branch, which may have changed since the harvest.

apply_rotary_pos_emb sail-sg/longspec/longspec/train/models/qwen2.py official repository ran · our draft was wrong MIT (permissive) · bac65c3dafaec040 · report
rotate_half sail-sg/longspec/longspec/train/models/qwen2.py official repository ran · our draft was wrong fingerprinted MIT (permissive) · b99eea6376d1e212 · report
build_dataloader sail-sg/longspec/longspec/train/general_util/evaluator.py official repository unverified MIT (permissive) · 8d0a17385afef041 · report
enable_gradient_checkpointing sail-sg/longspec/longspec/train/models/utils.py official repository unverified MIT (permissive) · a8d00b249de11718 · report
extract_and_rename sail-sg/longspec/longspec/train/trainer_base_ds_mul_fs_tp.py official repository unverified MIT (permissive) · bb5b320c91cd4e2d · report
forward_step sail-sg/longspec/longspec/train/trainer_base_ds_mul_fs_tp.py official repository unverified MIT (permissive) · 59492727dafc7ec5 · report
get_zero_stage sail-sg/longspec/longspec/train/trainer_base_ds_mul_fs_tp.py official repository unverified MIT (permissive) · c4393aecafc8b841 · report
maybe_contiguous sail-sg/longspec/longspec/train/models/triton_tree_attn.py official repository unverified MIT (permissive) · 58e6542897e7d424 · report
rounded_multiple sail-sg/longspec/longspec/train/models/triton_tree_attn.py official repository unverified MIT (permissive) · d0340dda6fca386f · report
torch_tree_attention sail-sg/longspec/longspec/train/models/qwen2.py official repository unverified MIT (permissive) · a8073503b1d41587 · report

Tasks

Code Completion

Results from the paper archive 2025-07-28

No leaderboard rows for this paper in the archive.

Methods

AttentionSoftmax

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections