metadata
license: apache-2.0
dataset_info:
features:
- name: smiles
dtype: string
- name: activity
dtype: float64
splits:
- name: train
num_bytes: 13736
num_examples: 513
- name: validation
num_bytes: 1721
num_examples: 64
- name: test
num_bytes: 1780
num_examples: 65
download_size: 15823
dataset_size: 17237
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: validation
path: data/validation-*
- split: test
path: data/test-*
The dataset is created using data from the deepchem library introduced by moleculenet. The dataset is split in a random fashion and normalized using z_score. The random seed used for splitting is 123. The smiles are canonicalized using RDkit.