Pith Number

pith:3KP6LBG6

pith:2026:3KP6LBG6FNOAWWGTSHCUNQHLU6

not attested not anchored not stored refs resolved

One Pass Is Not Enough: Recursive Latent Refinement for Generative Models

Alexia Jolicoeur-Martineau, Chirag Vashist, Ke Li, Mehdi Esmaeilzadeh

Replacing a single latent mapping with iterative refinement improves both image quality and diversity in generative models.

arxiv:2605.15309 v1 · 2026-05-14 · 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{3KP6LBG6FNOAWWGTSHCUNQHLU6}

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

Integrated with Implicit Maximum Likelihood Estimation (IMLE), RTM achieves the highest precision and recall among current state-of-the-art approaches while maintaining competitive FID, with improvements across CIFAR-10, CelebA-HQ at 256x256, and nine few-shot benchmarks.

C2weakest assumption

That performing multiple refinement iterations on the latent code will reliably increase mode coverage without destabilizing training or introducing new failure modes, an assumption invoked when the abstract states that recursive refinement improves both quality and diversity simultaneously.

C3one line summary

RTM uses iterative refinement of latent codes in generative models to improve both precision and recall alongside competitive FID scores on CIFAR-10, CelebA-HQ, and few-shot datasets.

References

12 extracted · 12 resolved · 6 Pith anchors

[1] Generative recursive reasoning models.ICLR 2026 Workshop on AI with Recursive Self-Improvement, 2026

[2] Mean Flows for One-step Generative Modeling · arXiv:2505.13447

[3] Goodfellow, Nips 2016 tutorial: Generative adversarial networks (2017), arXiv:1701.00160 [cs.LG] 2016 · arXiv:1701.00160

[4] Less is More: Recursive Reasoning with Tiny Networks · arXiv:2510.04871

[5] Implicit Maximum Likelihood Estimation · arXiv:1809.09087

Formal links

1 machine-checked theorem link

Receipt and verification

First computed	2026-05-20T00:00:51.898791Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

da9fe584de2b5c0b58d391c546c0eba7bcc3eb9f66c4597e0268257b795773f8

Aliases

arxiv: 2605.15309 · arxiv_version: 2605.15309v1 · doi: 10.48550/arxiv.2605.15309 · pith_short_12: 3KP6LBG6FNOA · pith_short_16: 3KP6LBG6FNOAWWGT · pith_short_8: 3KP6LBG6

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/3KP6LBG6FNOAWWGTSHCUNQHLU6 \
  | 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: da9fe584de2b5c0b58d391c546c0eba7bcc3eb9f66c4597e0268257b795773f8

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "f2a8e337b0b32921c728b072882a3bdd8f379610e9051875bcd5547ab393e20c",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-05-14T18:22:44Z",
    "title_canon_sha256": "9111f4abb7cdabc692f820878d03594739d7d0c51d2fce876e8da6d59e3602b6"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.15309",
    "kind": "arxiv",
    "version": 1
  }
}