AI Paper Atlas

Research map / Vision-language and multimodal / CLIP and zero-shot

PHyCLIP: $\ell_1$-Product of Hyperbolic Factors Unifies Hierarchy and Compositionality in Vision-Language Representation Learning

Daiki Yoshikawa, Takashi Matsubara

ICLR 2026 · Vision-language and multimodal › CLIP and zero-shot · approach: Recaptioning and text-supervised pretraining · 2 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use l1-product of hyperbolic factors (k=64, d=8) instead of a single hyperbolic space (ablation)HyCoCLIP · ImageNet zero-shottop-1 accuracy (%): 43.80 -> 44.31Improved
use l1-product of hyperbolic factors instead of a single hyperbolic space (ablation)HyCoCLIP · Food-101 zero-shottop-1 accuracy (%): 52.64 -> 57.26Improved
use l1-product of hyperbolic factors instead of a single hyperbolic space (ablation)HyCoCLIP · ImageNet hierarchical classificationTIE: 3.319 -> 3.294Improved
use l1 metric instead of l2-product metric (ablation)PHyCLIP · ImageNet zero-shottop-1 accuracy (%): 43.32 -> 44.31Improved
use l1 metric instead of l-infinity-product metric (ablation)PHyCLIP · ImageNet zero-shottop-1 accuracy (%): 6.55 -> 44.31Improved
use l1-product of hyperbolic factors instead of mixed-curvature space (ablation)PHyCLIP · ImageNet zero-shottop-1 accuracy (%): 39.34 -> 44.31Improved
use l1-product metric instead of mixed-curvature model (ablation)PHyCLIP · ImageNet hierarchical classificationTIE: 3.712 -> 3.294Improved
increase factors from 64 to 128 (d=4) (ablation)PHyCLIP · Food-101 zero-shottop-1 accuracy (%): 57.26 -> 53.96Hurt
increase factors from 8 to 64 (total dim fixed) (ablation)PHyCLIP · COCO text-to-image retrievalR@5: 57.80 -> 58.03Improved
increase factors from 64 to 128 (ablation)PHyCLIP · COCO image-to-text retrievalR@5: 70.94 -> 71.18Mixed
add box annotations to trainingCLIP · ImageNet zero-shottop-1 accuracy (%): 38.87 -> 38.81No clear effect
add box annotations to trainingMERU · ImageNet zero-shottop-1 accuracy (%): 37.96 -> 38.08No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy