Pith Number

pith:UI5FXM3M

pith:2026:UI5FXM3MUNOQWPVXHGKAVVWNLB

not attested not anchored not stored refs resolved

PipeSD: An Efficient Cloud-Edge Collaborative Pipeline Inference Framework with Speculative Decoding

Bing Hu, Mahdi Boloursaz Mashhadi, Pei Xiao, Yanfeng Zhang, Yitong Duan, Yunhe Han, Yunqi Gao

PipeSD speeds up cloud-edge LLM inference 1.16x-2.16x by pipelining token batches and flexible verification.

arxiv:2605.13319 v2 · 2026-05-13 · cs.DC

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{UI5FXM3MUNOQWPVXHGKAVVWNLB}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

PipeSD consistently outperforms state-of-the-art baselines, achieving 1.16x-2.16x speedup and reducing energy consumption by 14.3%-25.3%.

C2weakest assumption

The assumption that the dynamic-programming batch scheduler and Bayesian autotuner will deliver stable gains across unseen model pairs, network conditions, and workloads without introducing hidden overhead or requiring extensive per-deployment retuning.

C3one line summary

PipeSD achieves 1.16x-2.16x speedup and 14.3%-25.3% lower energy use in cloud-edge LLM inference via token-batch pipeline scheduling optimized by dynamic programming and a Bayesian-optimized dual-threshold NAV trigger.

References

70 extracted · 70 resolved · 0 Pith anchors

[1] Attribute-Based Bilateral Access Control With Sanitization and Trust Management for IIoT 2025 · doi:10.1109/tmc.2024.3513457

[2] Proceedings of the Workshop on Edge and Mobile Foundation Models , pages = 2024 · doi:10.1145/3662006.3662067

[3] EcoFed: Efficient Communication for DNN Partitioning-Based Federated Learning , year=

[4] Proceedings of the 14th USENIX Conference on Networked Systems Design and Implementation , pages = 2017

[5] Proceedings of the 40th International Conference on Machine Learning (ICML) , year =

Receipt and verification

First computed	2026-05-18T02:44:48.704293Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

a23a5bb36ca35d0b3eb739940ad6cd5854f247c4d1cbed056212da3060d1f9c1

Aliases

arxiv: 2605.13319 · arxiv_version: 2605.13319v2 · doi: 10.48550/arxiv.2605.13319 · pith_short_12: UI5FXM3MUNOQ · pith_short_16: UI5FXM3MUNOQWPVX · pith_short_8: UI5FXM3M

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/UI5FXM3MUNOQWPVXHGKAVVWNLB \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: a23a5bb36ca35d0b3eb739940ad6cd5854f247c4d1cbed056212da3060d1f9c1

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "abbb4ecbe13027990b416c3f1b2eb2f30160f0a3b8be4ed91f98cd7bad9024b3",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.DC",
    "submitted_at": "2026-05-13T10:34:04Z",
    "title_canon_sha256": "5b64317038cdaa9d3755a2da94d6bf59ad8183ba01293b76c0b361fedd34f55c"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.13319",
    "kind": "arxiv",
    "version": 2
  }
}