Datasets:

nielsr
/

datacomp-small-filtered

dataset_info:
  features:
    - name: uid
      dtype: string
    - name: url
      dtype: string
    - name: text
      dtype: string
    - name: original_width
      dtype: int64
    - name: original_height
      dtype: int64
    - name: clip_b32_similarity_score
      dtype: float32
    - name: clip_l14_similarity_score
      dtype: float32
    - name: face_bboxes
      sequence:
        sequence: float64
    - name: sha256
      dtype: string
    - name: clip_l14_embedding
      sequence: float64
  splits:
    - name: train
      num_bytes: 22741192622.844204
      num_examples: 3521953
  download_size: 6371420677
  dataset_size: 22741192622.844204

Dataset Card for "datacomp-small-filtered"

This is the DataComp-small dataset with CLIP-large-patch14 image embeddings added, as well as:

captions filtered for English using a FastText model
captions filtered to have at least complexity of 1