gte_embedded_movies / README.md
christophsonntag's picture
Update README.md
4d3370d verified
metadata
dataset_info:
  features:
    - name: runtime
      dtype: float64
    - name: languages
      sequence: string
    - name: metacritic
      dtype: float64
    - name: type
      dtype: string
    - name: imdb
      struct:
        - name: id
          dtype: int64
        - name: rating
          dtype: float64
        - name: votes
          dtype: int64
    - name: genres
      sequence: string
    - name: awards
      struct:
        - name: nominations
          dtype: int64
        - name: text
          dtype: string
        - name: wins
          dtype: int64
    - name: rated
      dtype: string
    - name: cast
      sequence: string
    - name: directors
      sequence: string
    - name: fullplot
      dtype: string
    - name: title
      dtype: string
    - name: num_mflix_comments
      dtype: int64
    - name: countries
      sequence: string
    - name: plot
      dtype: string
    - name: poster
      dtype: string
    - name: writers
      sequence: string
    - name: plot_embedding
      sequence: float64
  splits:
    - name: train
      num_bytes: 13657438
      num_examples: 1452
  download_size: 10625884
  dataset_size: 13657438
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

This dataset originates from MongoDB's embedded_movies dataset and contains details on movies from different genres. Each row represents a single movie with detailed information.

As opposed to the original dataset, this one includes embeddings of the fullplot column using the open source General Text Embeddings model instead of OpenAI's text-embedding-ada-002 embedding model used in MongoDB Atlas.

Those open source embeddings are also used in Hermes.