Pith Number

pith:UDCUC3AF

pith:2026:UDCUC3AFH3PF6AC2VWBVWKO6GD

not attested not anchored not stored refs resolved

Roll Out and Roll Back: Diffusion LLMs are Their Own Efficiency Teachers

Bo Han, Fanqin Zeng, Feng Hong, Geng Yu, Huangjie Zheng, Jiangchao Yao, Xiaofeng Cao, Yanfeng Wang, Ya Zhang

Diffusion LLMs discover reliable parallel decoding orders through revokable generation and then learn to use them for faster, higher-quality output.

arxiv:2605.16941 v1 · 2026-05-16 · cs.CL

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{UDCUC3AFH3PF6AC2VWBVWKO6GD}

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

DLLMs can serve as their own efficiency teachers by first discovering reliable denoising orders through revokable decoding and then learning to follow them for faster generation.

C2weakest assumption

The verification step in WINO, which uses enriched global context to decide which drafted tokens are reliable, correctly identifies tokens that will remain correct in the final output rather than merely appearing consistent at an intermediate step.

C3one line summary

Diffusion LLMs can act as their own efficiency teachers by using revokable parallel decoding to identify reliable token orders and then distilling those orders into the model parameters for faster inference.

References

45 extracted · 45 resolved · 11 Pith anchors

[1] Improving language understanding by generative pre-training, 2018

[2] Language models are unsupervised multitask learners, 2019

[3] Chatgpt: Optimizing language models for dialogue, 2022

[4] arXiv preprint arXiv:2310.12397 , year= 2023

[5] arXiv preprint arXiv:2310.08118 , year= 2023

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:03:32.036086Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

a0c5416c053ede5f005aad835b29de30d082004ffc9a9c4b42ed82f295b082cb

Aliases

arxiv: 2605.16941 · arxiv_version: 2605.16941v1 · doi: 10.48550/arxiv.2605.16941 · pith_short_12: UDCUC3AFH3PF · pith_short_16: UDCUC3AFH3PF6AC2 · pith_short_8: UDCUC3AF

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/UDCUC3AFH3PF6AC2VWBVWKO6GD \
  | 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: a0c5416c053ede5f005aad835b29de30d082004ffc9a9c4b42ed82f295b082cb

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "75fb7ce332bb2805d72fb72055c6f79c0676b0397551ef5ea973fe85c0b7f023",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CL",
    "submitted_at": "2026-05-16T11:27:40Z",
    "title_canon_sha256": "8797e6f37db35ef861c3e5d12809ce570b12adc869e71beb193184fc58ebb080"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.16941",
    "kind": "arxiv",
    "version": 1
  }
}