metadata
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: 636709030
num_examples: 75000
download_size: 503867836
dataset_size: 636709030
configs:
- config_name: default
data_files:
- split: train
path: data/train-*