metadata
dataset_info:
features:
- name: id
dtype: string
- name: dataset_source
dtype:
class_label:
names:
'0': hqedit
'1': omniedit
'2': ultraedit
- name: task
dtype: string
- name: input
dtype: image
- name: output
dtype: image
- name: instruction
dtype: string
- name: instruction_type
dtype:
class_label:
names:
'0': original
'1': rewrite
'2': complex
- name: input_description
dtype: string
- name: instruction_original
dtype: string
splits:
- name: train
num_bytes: 4410231338360.825
num_examples: 1553575
download_size: 4936041242856
dataset_size: 4410231338360.825
configs:
- config_name: default
data_files:
- split: train
path: data/train-*