Pith Number

pith:TZEAUCKG

pith:2026:TZEAUCKGSAI5WSHEJ3H5FIFJTE

not attested not anchored not stored refs resolved

Debunking Grad-ECLIP: A Comprehensive Study on Its Incorrectness and Fundamental Principles for Model Interpretation

Xiaohui Fan, Yongjin Cui

Grad-ECLIP produces model interpretations that do not match the original model's behavior or performance because its method is equivalent to a simpler attention-based route.

arxiv:2605.12952 v1 · 2026-05-13 · 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{TZEAUCKGSAI5WSHEJ3H5FIFJTE}

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

Both through formal derivation and experimental validation, we prove that the intermediate feature-based route represented by Grad-ECLIP is actually an equivalent variant of the attention-based route. ... the model interpretation results obtained by Grad-ECLIP are not those of the original model, and the interpretation results are misaligned with the model's performance.

C2weakest assumption

That the authors' Attention-ECLIP is exactly equivalent in all practical cases and that their experiments correctly isolate misalignment without selection bias or implementation differences.

C3one line summary

Grad-ECLIP is an equivalent but flawed variant of attention-based interpretation, with two principles proposed to ensure model explanations reflect the original model.

References

43 extracted · 43 resolved · 6 Pith anchors

[1] Quantifying attention flow in transformers 2020 · doi:10.18653/v1/2020.acl-main.385

[2] Deep integrated explanations 2023 · doi:10.1145/3583780.3614836

[3] Interpretability via model extrac- tion 2017 · arXiv:1706.09773

[4] Layer-wise relevance propagation for neural networks with local renormalization layers 2016 · doi:10.1007/978-3-319-44781-0

[5] Adabins: Depth estimation using adap- tive bins 2021 · doi:10.1109/cvpr46437.2021.00356

Receipt and verification

First computed	2026-05-18T03:09:09.380889Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

9e480a09469011db48e44ecfd2a0a9990fb54831bc01fbe6ae39f432d334566f

Aliases

arxiv: 2605.12952 · arxiv_version: 2605.12952v1 · doi: 10.48550/arxiv.2605.12952 · pith_short_12: TZEAUCKGSAI5 · pith_short_16: TZEAUCKGSAI5WSHE · pith_short_8: TZEAUCKG

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/TZEAUCKGSAI5WSHEJ3H5FIFJTE \
  | 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: 9e480a09469011db48e44ecfd2a0a9990fb54831bc01fbe6ae39f432d334566f

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "d0703a714d8e1ee51fceab37ba6ea4bb7f4954cc0c120c82d4a7e0a5b897a895",
    "cross_cats_sorted": [],
    "license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-05-13T03:35:23Z",
    "title_canon_sha256": "e66f3fa8d77000403b7a32642729898a947645124c215298b921aeb8f7eb84b1"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12952",
    "kind": "arxiv",
    "version": 1
  }
}