Universal Aesthetics (Multimodal Focus): Difference between revisions

From FDHwiki
Jump to navigation Jump to search
Line 5: Line 5:


== Data ==
== Data ==
As for the convergence of language models, we need both plain texts and aesthetic texts. For simplicity, we reuse this [https://huggingface.co/datasets/minhuh/prh/tree/wit_1024 text-image dataset], which is also used in Huh et al.'s paper<ref>{{cite arXiv
| title = The Platonic Representation Hypothesis
| author1 = Minyoung Huh
| author2 = Brian Cheung
| author3 = Tongzhou Wang
| author4 = Phillip Isola
| year = 2024
| eprint = 2405.07987
| archive-prefix = arXiv
| primary-class = cs.LG
| url = https://arxiv.org/abs/2405.07987
}}</ref>.
=== Plain Text ===


For poetry, we use the [https://www.kaggle.com/datasets/michaelarman/poemsdataset/data Poems dataset] from Kaggle, and categorize the poems into 135 types based on their form (haiku, sonnet, etc.).
=== Peoms ===
For poems, we use the [https://www.kaggle.com/datasets/michaelarman/poemsdataset/data Poems dataset] from Kaggle, and categorize the poems into 135 types based on their form (haiku, sonnet, etc.).
 
 
== References ==
<references />

Revision as of 21:17, 27 November 2025

Introduction

Methods

Data

As for the convergence of language models, we need both plain texts and aesthetic texts. For simplicity, we reuse this text-image dataset, which is also used in Huh et al.'s paper[1].

Plain Text

Peoms

For poems, we use the Poems dataset from Kaggle, and categorize the poems into 135 types based on their form (haiku, sonnet, etc.).


References