Pith Number

pith:VNFUHXCL

pith:2026:VNFUHXCLJAPFGIGBHHUNERJ47Y

not attested not anchored not stored refs resolved

HeatKV: Head-tuned KV-cache Compression for Visual Autoregressive Modeling

Axel Berg, Chuteng Zhou, Durmus Alp Emre Acar, Jonathan Cederlund, Pontus Giselsson

Head-specific attention ranking doubles KV-cache compression in visual autoregressive image models while preserving quality.

arxiv:2605.14877 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{VNFUHXCLJAPFGIGBHHUNERJ47Y}

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

Applied to the Infinity-2B model, HeatKV achieves 2× higher compression ratio in memory allocation for KV cache compared to existing methods, while maintaining similar or better image fidelity, prompt alignment and human perception score.

C2weakest assumption

That a static pruning schedule derived from attention scores on a small offline calibration set will generalize to arbitrary prompts and generation lengths without measurable quality loss.

C3one line summary

HeatKV ranks attention heads by their focus on prior scales using offline calibration data and applies a static per-head pruning schedule, delivering 2x higher KV-cache compression than prior methods on the Infinity-2B model with comparable image fidelity.

References

40 extracted · 40 resolved · 2 Pith anchors

[1] Infinity: Scaling bitwise autoregressive modeling for high-resolution image synthesis, 2025

[2] I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial nets,”Advances in neural information processing systems, vol. 27, 2014

[3] A style-based generator architecture for generative adversarial networks, 2019

[4] Stylegan-xl: Scaling stylegan to large diverse datasets, 2022

[5] Diffusion models beat gans on image synthesis, 2021

Formal links

2 machine-checked theorem links

Receipt and verification

First computed	2026-05-17T23:38:56.081842Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

ab4b43dc4b481e5320c139e8d2453cfe073be42803dbf4c2e4b9c684cad8d1ea

Aliases

arxiv: 2605.14877 · arxiv_version: 2605.14877v1 · doi: 10.48550/arxiv.2605.14877 · pith_short_12: VNFUHXCLJAPF · pith_short_16: VNFUHXCLJAPFGIGB · pith_short_8: VNFUHXCL

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/VNFUHXCLJAPFGIGBHHUNERJ47Y \
  | 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: ab4b43dc4b481e5320c139e8d2453cfe073be42803dbf4c2e4b9c684cad8d1ea

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "1ef38dabdc5d4752e11d311d9445b475b24b4c6ac345155aa2f530c37a22414c",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-05-14T14:22:34Z",
    "title_canon_sha256": "4b56ae184d7afff6d28f203935951a47bae3d761028d3320a9f5096242185147"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.14877",
    "kind": "arxiv",
    "version": 1
  }
}