---
dataset_info:
  features:
  - name: input
    dtype: string
  - name: output
    sequence: string
  splits:
  - name: train
    num_bytes: 411485
    num_examples: 1968
  - name: test
    num_bytes: 182161
    num_examples: 844
  download_size: 294736
  dataset_size: 593646
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---
This is a dataset of Amazon packaged food product reviews, annotated with food-related human values by design and psychology researchers from Stanford University.
Each data point is assigned to multiple value-emotion classes. This dataset can be used to train multi-label, multi-class classification models on product reviews.