metadata
dataset_info:
features:
- name: sentence1
dtype: string
- name: labels
dtype: string
- name: metadata
dtype: string
splits:
- name: train
num_bytes: 3938595
num_examples: 13595
- name: validation
num_bytes: 419503
num_examples: 1700
- name: test
num_bytes: 421194
num_examples: 1683
download_size: 933600
dataset_size: 4779292
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
license: cc-by-sa-4.0
source_datasets: []
streusle
STREUSLE native noun/verb supersenses, including discontinuous lexical expressions.
Original distribution: https://github.com/nert-nlp/streusle. Native UD train/dev/test and review IDs are retained. Adposition role/function distinctions are omitted from these noun/verb views. Dataset terms: CC BY-SA 4.0; original README records permission to redistribute source review sentences.
Original data: original distribution. Repackaged as parquet for tasksource by scripts/repackage_dataset/linguistic.py.
Conversion provenance: source revisions, hashes, configuration and exclusions.