pith:4VGU4GT2
When Do Diffusion Models learn to Generate Multiple Objects?
Diffusion models' multi-object generation is limited primarily by scene complexity and held-out combinations rather than imbalance, with counting difficult in low data and compositional generalization collapsing as more combinations are excluded.
arxiv:2605.00273 v2 · 2026-04-30 · cs.CV · cs.AI
Add to your LaTeX paper
\usepackage{pith}
\pithnumber{4VGU4GT2VH2UJ62QWMHBW3SFBK}
Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge
Record completeness
Claims
By training diffusion models on mosaic, we find that scene complexity plays a dominant role rather than concept imbalance, and that counting is uniquely difficult to learn in low-data regimes. Moreover, compositional generalization collapses as more concept combinations are held out during training.
That the synthetic MOSAIC datasets and the defined regimes of concept versus compositional generalization capture the essential factors driving failures in real-world text-to-image diffusion models trained on natural image distributions.
Diffusion models' multi-object generation is limited primarily by scene complexity and held-out combinations rather than imbalance, with counting difficult in low data and compositional generalization collapsing as more combinations are excluded.
Receipt and verification
| First computed | 2026-06-09T02:08:43.167864Z |
|---|---|
| Builder | pith-number-builder-2026-05-17-v1 |
| Signature | Pith Ed25519
(pith-v1-2026-05) · public key |
| Schema | pith-number/v1.0 |
Canonical hash
e54d4e1a7aa9f544fb50b30e1b6e450a9c99bbca5ebd078103d5d9295c49ae52
Aliases
· · · · ·Agent API
Verify this Pith Number yourself
curl -sH 'Accept: application/ld+json' https://pith.science/pith/4VGU4GT2VH2UJ62QWMHBW3SFBK \
| 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: e54d4e1a7aa9f544fb50b30e1b6e450a9c99bbca5ebd078103d5d9295c49ae52
Canonical record JSON
{
"metadata": {
"abstract_canon_sha256": "62e90c1b7550aaf042a0b066dcb80322649345bf39790d22c032f0eabbd4c047",
"cross_cats_sorted": [
"cs.AI"
],
"license": "http://creativecommons.org/licenses/by/4.0/",
"primary_cat": "cs.CV",
"submitted_at": "2026-04-30T22:18:33Z",
"title_canon_sha256": "f881bbc380d7acf33c862467119c0bc3d776c56632d851eb0afbae4bae64ad09"
},
"schema_version": "1.0",
"source": {
"id": "2605.00273",
"kind": "arxiv",
"version": 2
}
}