Fast Data Attribution for Text-to-Image Models
A preprint pinned to a corkboard: plain grey Helvetica, blue bracketed links, and the plots doing all the talking.
Checked against the rendered page. No web fonts load; on Windows the Helvetica Neue Light stack renders as regular-weight Arial.
Why it is here
A spare, almost paper-like style with plain serif-free type and thin rules. It adds variety by putting large, legible result plots (a scatter with speedup arrows, a bar chart) at the centre of the page instead of decoration.
Overview
Color Palette
Typography
Helvetica Neue Light
Fast Data Attribution for Text-to-Image Models
Abstract · Method · Results · BibTeX — 0123456789
Arial
Fast Data Attribution for Text-to-Image Models
Abstract · Method · Results · BibTeX — 0123456789
Spacing & Shape
Page Modules
Guidelines
More like this
CRAFT: Training-Free Cascaded Retrieval for Tabular QA
A tidy product manual in pale blueprint blue, where every stage wears its own colored edge.
Efficient Self-Evaluation for Diffusion Language Models via Sequence Regeneration
A tidy stack of rounded cards washed in a blue-violet-apricot dawn, signed with a gradient wordmark.
FIGMA: Towards FIne-Grained Music retrievAl
A music journal on warm paper: a heavy Newsreader headline under a red kicker, big red serif numerals for the results, and figures resting in soft cream frames.
GuideDog: A Real-World Egocentric Multimodal Dataset for Blind and Low-Vision Accessibility-Aware Guidance
A quiet monograph page: one serif column, oxblood ink for the links, and a table that blushes only at its own row.
LaMI: Augmenting Large Language Models via Late Multi-Image Fusion
A careful conference handout: Noto Serif headings, navy outlined buttons and soft grey bands taking turns.
PairCoder++: Pair Programming as a Universal Paradigm for Verified Code-Driven Multimodal and Structured-Artifact Generation
A crisp white product page: a blue announcement strip, ink-black pill buttons and every figure on a soft-shadowed card.