metadata
dataset_info:
features:
- name: premise
dtype: string
- name: hypothesis
dtype: string
- name: label
dtype:
class_label:
names:
'0': entailment
'1': neutral
'2': contradiction
- name: input_ids
sequence: int32
- name: attention_mask
sequence: int8
- name: token_type_ids
sequence: int8
splits:
- name: train
num_bytes: 3705841865
num_examples: 5890530
- name: test
num_bytes: 45951784
num_examples: 75150
- name: validation
num_bytes: 22624609
num_examples: 37350
download_size: 2055809923
dataset_size: 3774418258
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: validation
path: data/validation-*