metadata
dataset_info:
features:
- name: document name
dtype: string
- name: tokens
sequence: string
- name: tokens-IDs
sequence: int8
- name: ner_tags
sequence: string
- name: sentence-IDs
sequence: int8
- name: relations
sequence:
- name: source-head-sentence-ID
dtype: int8
- name: source-head-word-ID
dtype: int8
- name: relation-type
dtype: string
- name: target-head-sentence-ID
dtype: int8
- name: target-head-word-ID
dtype: int8
splits:
- name: test
num_bytes: 203379
num_examples: 45
download_size: 38326
dataset_size: 203379
This is the version 1.1.0 of the original PET dataset. in this version we fixed ``the Performs Relations'' and few minor errors.
Please refer to the original PET Dataset repository for more info.