metadata
dataset_info:
features:
- name: id
dtype: string
- name: conversations
list:
- name: from
dtype: string
- name: value
dtype: string
- name: ram++_tags
sequence: string
- name: ram++_tags_score
sequence: float64
- name: image_tokens
sequence: int64
- name: image_shape
sequence: int64
splits:
- name: train
num_bytes: 4676645851
num_examples: 3557630
download_size: 2176970066
dataset_size: 4676645851
configs:
- config_name: default
data_files:
- split: train
path: data/train-*