metadata
dataset_info:
features:
- name: sequence
dtype: string
- name: label
dtype: int64
splits:
- name: train
num_bytes: 2893542
num_examples: 3424
- name: valid
num_bytes: 740946
num_examples: 857
- name: test
num_bytes: 2594652
num_examples: 2529
download_size: 2914341
dataset_size: 6229140
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: valid
path: data/valid-*
- split: test
path: data/test-*
From this paper. 1 is high expression, 0 is low.