---
dataset_info:
- config_name: 100k
  features:
  - name: key
    dtype: string
  - name: clip_similarity_vitb32
    dtype: float64
  - name: clip_similarity_vitl14
    dtype: float64
  - name: id
    dtype: int64
  - name: url
    dtype: string
  - name: caption
    dtype: string
  - name: width
    dtype: int64
  - name: height
    dtype: int64
  - name: noun_chunks
    sequence:
      sequence: float64
  - name: ref_exps
    sequence:
      sequence: float64
  - name: image
    dtype: image
  splits:
  - name: train
    num_bytes: 11306203943.7756
    num_examples: 151428
  download_size: 14833282357
  dataset_size: 11306203943.7756
- config_name: 1M_0
  features:
  - name: key
    dtype: string
  - name: clip_similarity_vitb32
    dtype: float64
  - name: clip_similarity_vitl14
    dtype: float64
  - name: id
    dtype: int64
  - name: url
    dtype: string
  - name: caption
    dtype: string
  - name: width
    dtype: int64
  - name: height
    dtype: int64
  - name: noun_chunks
    sequence:
      sequence: float64
  - name: ref_exps
    sequence:
      sequence: float64
  - name: image
    dtype: image
  splits:
  - name: train
    num_bytes: 74014470938.25
    num_examples: 757798
  download_size: 73664367108
  dataset_size: 74014470938.25
- config_name: 47k
  features:
  - name: key
    dtype: string
  - name: clip_similarity_vitb32
    dtype: float64
  - name: clip_similarity_vitl14
    dtype: float64
  - name: id
    dtype: int64
  - name: url
    dtype: string
  - name: caption
    dtype: string
  - name: width
    dtype: int64
  - name: height
    dtype: int64
  - name: noun_chunks
    sequence:
      sequence: float64
  - name: ref_exps
    sequence:
      sequence: float64
  - name: image
    dtype: image
  splits:
  - name: train
    num_bytes: 4596304378.22
    num_examples: 47506
  download_size: 4569436913
  dataset_size: 4596304378.22
configs:
- config_name: 100k
  data_files:
  - split: train
    path: 100k/train-*
- config_name: 1M_0
  data_files:
  - split: train
    path: 1M_0/train-*
- config_name: 1M_grit_0
  data_files:
  - split: train
    path: 1M_grit_0/train-*
- config_name: 47k
  data_files:
  - split: train
    path: 47k/train-*
---