metadata
dataset_info:
features:
- name: image
dtype: image
- name: unicode
dtype: string
- name: meaning
sequence: string
- name: primary_meaning
dtype: string
- name: prompt
dtype: string
splits:
- name: train
num_bytes: 31894933
num_examples: 6410
download_size: 32122546
dataset_size: 31894933
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
Generated from Kanjivg. 128 x 128 images with meanings. https://kanjivg.tagaini.net/index.html