diffusiondb_ner / README.md
teo-sanchez
readme.md update
d05d1b5
metadata
layout: default
title: Name Entity Recognition of DiffusionDB
nav_order: 1
has_children: false
language_creators:
  - found
language:
  - en
license:
  - cc-by-3.0
multilinguality:
  - monolingual
pretty_name: NER-DiffusionDB
size_categories:
  - 100M<n<1G
source_datasets:
  - poloclub/diffusiondb
tags:
  - stable diffusion
  - prompt engineering
  - prompts
  - research paper

Description

Extended dataset infered by the name entity recognition model en_ner_prompting. This model has been trained on hand-annotated prompts from poloclub/diffusiondb. This dataset is hence infered by this model and can comprise mistakes, especially on certain categories (cf. model card).

The entities comprise 7 main categories and 11 subcategories for a total of 16 categories, extracted from a topic analysis made with BERTopic. The topic analysis can be explored the following visualization.

  ├── medium/
  │   ├── photography
  │   ├── painting
  │   ├── rendering
  │   └── illustration
  ├── influence/
  │   ├── artist
  │   ├── genre
  │   ├── artwork
  │   └── repository
  ├── light
  ├── color
  ├── composition
  ├── detail
  └── context/
      ├── era
      ├── weather
      └── emotion

Label Scheme

View label scheme (16 labels for 1 components)
Component Labels
ner color, composition, context/emotion, context/era, context/weather, detail, influence/artist, influence/artwork, influence/genre, influence/repository, light, medium/illustration, medium/painting, medium/photography, medium/rendering, subject