Pith Number

pith:JQJ2RQJB

pith:2026:JQJ2RQJBQYASI4AKW3GQ2NIWQS

not attested not anchored not stored refs resolved

CAR-SAM: Cross-Attention Reconstruction for Post-Training Quantization of the Segment Anything Model

Dawei Yang, Houji Wen, Jiangyong Yu, Jun Li

CAR-SAM enables effective 4-bit post-training quantization of the Segment Anything Model by fixing decoder attention issues.

arxiv:2605.16901 v1 · 2026-05-16 · 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{JQJ2RQJBQYASI4AKW3GQ2NIWQS}

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

CAR-SAM robustly quantizes SAM models down to 4-bit precision, surpassing existing methods by 14.6% and 6.6% mAP on SAM-B and SAM-L respectively.

C2weakest assumption

The primary degradation in existing PTQ for SAM stems from attention dissipation and reconstruction oscillation in the decoder, and that MAC and JCAR mechanisms directly mitigate these without introducing new instabilities or requiring model-specific retraining.

C3one line summary

CAR-SAM introduces MatMul-Aware Compensation and Joint Cross-Attention Reconstruction to enable stable 4-bit post-training quantization of SAM, outperforming prior PTQ methods by 14.6% mAP on SAM-B and 6.6% on SAM-L.

References

29 extracted · 29 resolved · 4 Pith anchors

[1] Lsq+: Improving low-bit quantization through learnable offsets and better initializa- tion 2020

[2] Slimsam: 0.1% data makes segment anything slim 2024

[3] Learned step size quantization 1902

[4] GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers 2022 · arXiv:2210.17323

[5] Segment any- thing 2023

Receipt and verification

First computed	2026-05-20T00:03:29.167241Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

4c13a8c121860124700ab6cd0d3516848bf989482cf23c8551c38ce6629b4820

Aliases

arxiv: 2605.16901 · arxiv_version: 2605.16901v1 · doi: 10.48550/arxiv.2605.16901 · pith_short_12: JQJ2RQJBQYAS · pith_short_16: JQJ2RQJBQYASI4AK · pith_short_8: JQJ2RQJB

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/JQJ2RQJBQYASI4AKW3GQ2NIWQS \
  | 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: 4c13a8c121860124700ab6cd0d3516848bf989482cf23c8551c38ce6629b4820

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "852671bb7aa0c6dcc2ce53073a454b1de263183adede5685429d29f084749701",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-05-16T09:25:23Z",
    "title_canon_sha256": "50f104fd5abe006b6c103c9f7333854c3b5bcb5cc1b71a03c8e54b93cfe8eb47"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.16901",
    "kind": "arxiv",
    "version": 1
  }
}