Metraipolis › archive › dc_d72df13089b7

TRACE: Transparent Retrieval for Abstract Concept Evaluation — arXiv:2609.26168

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.26168
License (as archived)
CC0-1.0
Archived
2026-10-08T15:06:57.955Z
SHA-256 of the archived text
dbf68d1d2f8c68b50446bc19684517698e97d3805ccb7e04e9daba72ee8de5cc
Stable URL of this page
https://metraipolis.com/documento/dc_d72df13089b7

How to cite

TRACE: Transparent Retrieval for Abstract Concept Evaluation — arXiv:2609.26168 — arXiv (en), https://arxiv.org/abs/2609.26168 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_d72df13089b7 (Metraipolis archive dc_d72df13089b7, sha256 dbf68d1d2f8c68b50446bc19684517698e97d3805ccb7e04e9daba72ee8de5cc)
BibTeX
@misc{dcd72df13089b7,
  title = {TRACE: Transparent Retrieval for Abstract Concept Evaluation — arXiv:2609.26168},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.26168},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_d72df13089b7; sha256 dbf68d1d2f8c68b50446bc19684517698e97d3805ccb7e04e9daba72ee8de5cc},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.26168. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Recent work reports that vision--language models (VLMs) struggle to establish and maintain stable reference in repeated reference games. Rather than ask which VLM does best, we ask a more basic question: do you need a large pretrained VLM for this at all? On grounding a single director utterance to one of twelve tangram silhouettes, we compare six off-the-shelf VLMs against a transparent baseline that uses \emph{no learned visual representation}: classical SIFT keypoint matching and a signal-quality index over retrieved images. On identical trials, the transparent baseline matches the strongest VLM (SigLIP-large) and significantly outperforms the other five, including every CLIP and OpenCLIP variant. The baseline additionally retrieves external images, so this is not a matched-information comparison; what it shows is that a learned \emph{visual} representation is not the bottleneck for this task: given retrieved images, a shape-appropriate classical similarity suffices. Along the way we find that abstract-grounding ability varies widely across VLMs (15--39\% top-1; chance 8.33\%, humans $\approx$77--80\%), so the weakness is model-specific rather than intrinsic to contrastive pretraining; on the 1{,}013-shape KiloGram benchmark the pattern generalizes for CLIP, with per-shape difficulty tracking human shape-nameability. The pipeline is a classical, inspectable alternative rather than a learned one. We close by sketching how an explicit, inspectable representation of listener-side pact state could carry this approach into interactive multi-turn reference, which we leave to future work. Code available in supplementary material.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-23 in cs.HC).
Categorie arXiv: Human-Computer Interaction (cs.HC) ; Computer Vision and Pattern Recognition (cs.CV).
Pagina del lavoro: https://arxiv.org/abs/2609.26168 — 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 2026-10-08T14:34:31.474Z dalla pagina https://arxiv.org/catchup/cs.HC/2026-09-23?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).