数据结构如下：

dataset_info:
  features:
  - name: id
    dtype: string
  - name: title
    dtype: string
  - name: short_description
    dtype: string
  - name: img_high
    dtype: string
  - name: supplier
    dtype: string
  - name: text
    dtype: string
  - name: embeddings
    sequence: float32
  splits:
  - name: train
    num_bytes: 65690182
    num_examples: 19406
  download_size: 75241335
  dataset_size: 65690182