Metraipolis › archive › dc_fb306f892adb
Influence Score and Transformers interpretability: Measure of the Effective Impact of Attention Heads at inference time — arXiv:2609.05074
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.05074
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-16T22:35:25.829Z
- SHA-256 of the archived text
0b92810c896c04a0dc248676cf761cefba4d0369080ec9422c64b0cceec764aa- Stable URL of this page
https://metraipolis.com/documento/dc_fb306f892adb
How to cite
Influence Score and Transformers interpretability: Measure of the Effective Impact of Attention Heads at inference time — arXiv:2609.05074 — arXiv (en), https://arxiv.org/abs/2609.05074 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_fb306f892adb (Metraipolis archive dc_fb306f892adb, sha256 0b92810c896c04a0dc248676cf761cefba4d0369080ec9422c64b0cceec764aa)
BibTeX
@misc{dcfb306f892adb,
title = {Influence Score and Transformers interpretability: Measure of the Effective Impact of Attention Heads at inference time — arXiv:2609.05074},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.05074},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_fb306f892adb; sha256 0b92810c896c04a0dc248676cf761cefba4d0369080ec9422c64b0cceec764aa},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.05074. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We propose an influence score to quantify the contribution of attention heads to classification decisions in Transformer-based models designed for prompt injection detection. The score combines directional influence on the logits with structural contribution within the residual stream, enabling a multi-scale analysis at the head, layer, and network levels. Applied to a DeBERTa model specialized for prompt injection detection, our framework reveals distinct decision behaviours between correct and erroneous predictions. Our method provides an effective compromise between fine-grained circuit analysis and global output-based methods, and offers a systematic way to study decision mechanisms in Transformer classifiers. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-07 in cs.CL). Categorie arXiv: Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2609.05074 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-16T22:21:16.293Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-07?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).