metadata
dataset_info:
features:
- name: doc_id
dtype: string
- name: text
dtype: string
- name: annotations
list:
- name: confidence
dtype: float64
- name: end
dtype: int64
- name: field
dtype: string
- name: is_null
dtype: bool
- name: loss_weight
dtype: float64
- name: span_text
dtype: string
- name: start
dtype: int64
- name: value
dtype: string
- name: schema_type
dtype: string
splits:
- name: train
num_bytes: 34569869
num_examples: 15671
- name: holdout
num_bytes: 878903
num_examples: 301
download_size: 11367571
dataset_size: 35448772
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: holdout
path: data/holdout-*