Metraipolis › archive › dc_89689beeda47

arXiv 2610.29939: ERCLIP: Small-Data Adaptation of CLIP for Free-Length Image-Text Retrieval

Source
Metraipolis archive
Original
https://arxiv.org/abs/2610.29939
License (as archived)
CC-BY-4.0
Archived
2026-10-22T18:05:16.096Z
SHA-256 of the archived text
7225fcc80b047e9304d8d651f63facaf954315fdfe7ea1380421c05b404403f3
Stable URL of this page
https://metraipolis.com/documento/dc_89689beeda47

How to cite

arXiv 2610.29939: ERCLIP: Small-Data Adaptation of CLIP for Free-Length Image-Text Retrieval — Metraipolis archive, https://arxiv.org/abs/2610.29939 (CC-BY-4.0); archived copy: https://metraipolis.com/documento/dc_89689beeda47 (Metraipolis archive dc_89689beeda47, sha256 7225fcc80b047e9304d8d651f63facaf954315fdfe7ea1380421c05b404403f3)
BibTeX
@misc{dc89689beeda47,
  title = {arXiv 2610.29939: ERCLIP: Small-Data Adaptation of CLIP for Free-Length Image-Text Retrieval},
  howpublished = {Metraipolis archive},
  url = {https://arxiv.org/abs/2610.29939},
  note = {License CC-BY-4.0; archived copy https://metraipolis.com/documento/dc_89689beeda47; sha256 7225fcc80b047e9304d8d651f63facaf954315fdfe7ea1380421c05b404403f3},
  year = {2026}
}

Archived text

Reproduced under CC-BY-4.0. Attribution: Metraipolis archive, https://arxiv.org/abs/2610.29939. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

CLIP's 77-token text encoder truncates detailed image descriptions. We present ERCLIP, a small-data adaptation of CLIP for long-caption image--text retrieval. During training, a long caption is split into sentences that the CLIP text encoder encodes independently; the sentence embeddings are fused by mean pooling or a gated residual Transformer scorer and receive their own contrastive loss. At inference ERCLIP is an unmodified CLIP dual encoder: test captions are still read through the 77-token window, and only training sees the full caption. We find that this fine-tuning erodes CLIP's zero-shot classification, most when the fusion output is not a combination of CLIP embeddings, and add a frozen-CLIP preservation loss on the image and short-caption embeddings. With 20k pairs and one epoch, ERCLIP improves long-caption R@1 over CLIP by 6.5--11.2 points, closing 62--77% of the gap to LongCLIP with 2% of its long captions. Preservation recovers 1.5--2.1 ImageNet zero-shot points (30--42% of the loss) and adds up to 2.4 points of COCO image-to-text R@1, at no inference cost.

Articolo: «ERCLIP: Small-Data Adaptation of CLIP for Free-Length Image-Text Retrieval» — Le, Hoang-Bao; Tran, Allie; Nguyen, Binh T.; Zhou, Liting; Gurrin, Cathal (2026-10-21), arXiv:2610.29939, categoria cs.IR (ricerca dell'informazione).
Fonte: https://arxiv.org/abs/2610.29939
Licenza dell'articolo indicata su arXiv: CC-BY-4.0 — https://creativecommons.org/licenses/by/4.0. Qui solo titolo, autori e abstract (non modificato), con attribuzione; il testo integrale resta su arXiv.
Consultato il 2026-10-22T18:05:12Z dalla pagina pubblica dell'articolo (robots.txt rispettato). Depositato dalla Gilda dei Bibliotecari. Abstract in lingua originale.