Pith Number

pith:Z3SH3GDB

pith:2024:Z3SH3GDBVKBL4C5BHZ2THBUHEM

not attested not anchored not stored refs resolved

One Step Diffusion via Shortcut Models

Danijar Hafner, Kevin Frans, Pieter Abbeel, Sergey Levine

Shortcut models generate high-quality diffusion samples in one step using a single network.

arxiv:2410.12557 v3 · 2024-10-16 · cs.LG · cs.CV

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

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{Z3SH3GDBVKBL4C5BHZ2THBUHEM}

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

Shortcut models consistently produce higher quality samples than previous approaches, such as consistency models and reflow. Compared to distillation, shortcut models reduce complexity to a single network and training phase and additionally allow varying step budgets at inference time.

C2weakest assumption

That a single network can learn effective large-step transitions across a wide range of step sizes during one training phase without quality degradation or the need for fragile scheduling.

C3one line summary

Shortcut models enable high-quality single or few-step sampling in diffusion models with one network and training phase by conditioning on desired step size.

References

28 extracted · 28 resolved · 14 Pith anchors

[1] Lumiere: A space-time diffusion model for video generation

[2] Tract: Denoising diffusion models with transitive closure time-distillation

[3] arXiv preprint arXiv:2406.07507 (2024) 5

[4] Large Scale GAN Training for High Fidelity Natural Image Synthesis · arXiv:1809.11096

[5] Diffusion Policy: Visuomotor Policy Learning via Action Diffusion · arXiv:2303.04137

Formal links

3 machine-checked theorem links

Cited by

33 papers in Pith

Lipschitz-Guided Design of Interpolation Schedules in Generative Models

Flow Map Language Models: One-step Language Modeling via Continuous Denoising

How to Guide Your Flow: Few-Step Alignment via Flow Map Reward Guidance

Efficient Image Synthesis with Sphere Latent Encoder

Stabilizing, Scaling & Enhancing MeanFlow for Large-scale Diffusion Distillation

Receipt and verification

First computed	2026-05-17T23:38:53.233683Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

cee47d9861aa82be0ba13e7533868723279b108f804c8eb8e2e98fc581828d2f

Aliases

arxiv: 2410.12557 · arxiv_version: 2410.12557v3 · doi: 10.48550/arxiv.2410.12557 · pith_short_12: Z3SH3GDBVKBL · pith_short_16: Z3SH3GDBVKBL4C5B · pith_short_8: Z3SH3GDB

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/Z3SH3GDBVKBL4C5BHZ2THBUHEM \
  | 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: cee47d9861aa82be0ba13e7533868723279b108f804c8eb8e2e98fc581828d2f

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "5d0dbb8d6952cd56a5ec73a36c02b8cb6fd2a48bd9091654cd6e00a68e10c8b9",
    "cross_cats_sorted": [
      "cs.CV"
    ],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.LG",
    "submitted_at": "2024-10-16T13:34:40Z",
    "title_canon_sha256": "3fc05110424b39f5ffe0b26a1dc0e3e541fa5acf9c2d6d3a031b40b293d7abc5"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2410.12557",
    "kind": "arxiv",
    "version": 3
  }
}