Pith Number

pith:KFKTDL5H

pith:2026:KFKTDL5H6ZCIZBOTZJ73ZEO5UE

not attested not anchored not stored refs resolved

DMax: Aggressive Parallel Decoding for dLLMs

Gongfan Fang, Ruonan Yu, Xinchao Wang, Xinyin Ma, Zigeng Chen

DMax reformulates parallel decoding for diffusion language models as progressive self-refinement from mask to token embeddings.

arxiv:2604.08302 v3 · 2026-04-09 · cs.LG · cs.AI

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{KFKTDL5H6ZCIZBOTZJ73ZEO5UE}

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

Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance.

C2weakest assumption

That representing each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding enables effective iterative self-revising in embedding space without accumulating new errors or requiring additional post-processing.

C3one line summary

DMax uses On-Policy Uniform Training and Soft Parallel Decoding to enable aggressive parallelism in dLLMs, raising TPF on GSM8K from 2.04 to 5.47 and on MBPP from 2.71 to 5.86 while preserving accuracy.

References

110 extracted · 110 resolved · 24 Pith anchors

[1] GPT-4 Technical Report 2023 · arXiv:2303.08774

[2] Opencodeinstruct: A large-scale instruction tuning dataset for code llms 2025

[3] Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models 2025 · arXiv:2503.09573

[4] Structured denoising diffusion models in discrete state-spaces.Advances in neural information processing systems, 34:17981–17993 2021

[5] Program Synthesis with Large Language Models 2021 · arXiv:2108.07732

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-20T00:00:37.787996Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c

Aliases

arxiv: 2604.08302 · arxiv_version: 2604.08302v3 · doi: 10.48550/arxiv.2604.08302 · pith_short_12: KFKTDL5H6ZCI · pith_short_16: KFKTDL5H6ZCIZBOT · pith_short_8: KFKTDL5H

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/KFKTDL5H6ZCIZBOTZJ73ZEO5UE \
  | 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: 515531afa7f6448c85d3ca7fbc91dda137e8fdf76d1b98e180ad1c4ef4b31d6c

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "b2fbbf793f6564d8d824c34f8ff318f7b8e9dcc70d0a8b7f72eb494ac28afed4",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-04-09T14:35:42Z",
    "title_canon_sha256": "bb2f046bd6663f117409a9fd6365c2e74dc4bd6fa03d41f06e4e26c85543b725"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2604.08302",
    "kind": "arxiv",
    "version": 3
  }
}