---
dataset_info:
  features:
  - name: instruction_seed
    dtype: string
  - name: source
    dtype: string
  - name: embeddings
    sequence: float64
  - name: mean_positive_score
    dtype: float64
  - name: mean_negative_score
    dtype: float64
  - name: difference_score
    dtype: float64
  splits:
  - name: train
    num_bytes: 37218369
    num_examples: 4000
  download_size: 28045681
  dataset_size: 37218369
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
---