metadata
dataset_info:
features:
- name: id
dtype: string
- name: request
dtype: string
- name: response
dtype: string
- name: input_perplexity
dtype: float64
- name: input_likelihood
dtype: float64
- name: output_perplexity
dtype: float64
- name: output_likelihood
dtype: float64
splits:
- name: validation
num_bytes: 2716450
num_examples: 3270
download_size: 1480464
dataset_size: 2716450
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*