Pith Number

pith:NTWZA4UI

pith:2026:NTWZA4UIH5XFDMAU3BV2QNB3OI

not attested not anchored not stored refs resolved

Orthrus: Memory-Efficient Parallel Token Generation via Dual-View Diffusion

Chaitra Hegde, Chien Van Nguyen, Franck Dernoncourt, Ryan A. Rossi, Thien Huu Nguyen, Van Cuong Pham

Orthrus adds a lightweight diffusion view to frozen LLMs so they can generate tokens in parallel while matching standard autoregressive output exactly.

arxiv:2605.12825 v1 · 2026-05-12 · 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{NTWZA4UIH5XFDMAU3BV2QNB3OI}

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

By employing an exact consensus mechanism between the two views, Orthrus guarantees lossless inference, delivering up to a 7.8x speedup with only an O(1) memory cache overhead and minimal parameter additions.

C2weakest assumption

The exact consensus mechanism between the autoregressive and diffusion views will always produce identical output to standard autoregressive decoding without introducing any quality degradation or requiring additional training data.

C3one line summary

Orthrus unifies autoregressive and diffusion views on a shared KV cache to deliver lossless parallel token generation with up to 7.8x speedup and O(1) memory overhead.

References

25 extracted · 25 resolved · 14 Pith anchors

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

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

[3] Program Synthesis with Large Language Models 2026 · arXiv:2108.07732

[4] Language models are few-shot learners.Advances in neural information processing systems, 33:1877–1901 1901

[5] Zhuoming Chen, Avner May, Ruslan Svirschevski, Yuhsun Huang, Max Ryabinin, Zhihao Jia, and Beidi Chen

Receipt and verification

First computed	2026-05-18T03:09:12.165558Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

6ced9072883f6e51b014d86ba8343b721440c455dacebe62c1984fdc7475b308

Aliases

arxiv: 2605.12825 · arxiv_version: 2605.12825v1 · doi: 10.48550/arxiv.2605.12825 · pith_short_12: NTWZA4UIH5XF · pith_short_16: NTWZA4UIH5XFDMAU · pith_short_8: NTWZA4UI

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/NTWZA4UIH5XFDMAU3BV2QNB3OI \
  | 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: 6ced9072883f6e51b014d86ba8343b721440c455dacebe62c1984fdc7475b308

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "a55b2eafd4e5b857c91d8247ac615ad8f7469d957024becaad6dc1365e4203a2",
    "cross_cats_sorted": [
      "cs.AI"
    ],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2026-05-12T23:47:35Z",
    "title_canon_sha256": "05da86b68e5d5d3b1f60a664c9b3d780981d51cb2252d88d47fccda2f5af9d48"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12825",
    "kind": "arxiv",
    "version": 1
  }
}