metadata
dataset_info:
features:
- name: index
dtype: string
- name: url
dtype: string
- name: raw_prompt
dtype: string
- name: ground_truth
dtype: image
- name: gen_images_sd1_4
list: image
- name: minimal_rephrased_prompt
dtype: string
- name: gen_images_minimal_rephrased
list: image
- name: mse_minimal_rephrased
dtype: float32
- name: gpt_rephrased_prompt
dtype: string
- name: gen_images_gpt_rephrased
list: image
- name: mse_gpt_rephrased
dtype: float32
- name: qwen_recaptioned_prompt
dtype: string
- name: gen_images_qwen_recaptioned
list: image
- name: mse_qwen_recaptioned
dtype: float32
splits:
- name: train
num_bytes: 3036242326
num_examples: 500
download_size: 3023380977
dataset_size: 3036242326
configs:
- config_name: default
data_files:
- split: train
path: data/train-*