metadata
dataset_info:
features:
- name: original_input
dtype: string
- name: input_text
dtype: string
- name: output
dtype: string
- name: spurious_feature
dtype: string
splits:
- name: train
num_bytes: 2224143
num_examples: 6620
- name: test_iid
num_bytes: 608768
num_examples: 1818
- name: test_high
num_bytes: 604943
num_examples: 1818
- name: test_low
num_bytes: 606891
num_examples: 1818
- name: test_flipped
num_bytes: 607351
num_examples: 1818
download_size: 2755922
dataset_size: 4652096
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test_iid
path: data/test_iid-*
- split: test_high
path: data/test_high-*
- split: test_low
path: data/test_low-*
- split: test_flipped
path: data/test_flipped-*