metadata
dataset_info:
features:
- name: smiles
dtype: string
- name: logp
dtype: float64
- name: sascore
dtype: float64
- name: mol_weight
dtype: float64
splits:
- name: train
num_bytes: 749464427.0832263
num_examples: 10388168
- name: test
num_bytes: 93683125.53137042
num_examples: 1298522
- name: valid
num_bytes: 93683053.38540329
num_examples: 1298521
download_size: 705437904
dataset_size: 936830606
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
- split: test
path: data/test-*
- split: valid
path: data/valid-*
license: mit
dataset description
Following the instructions in LlaMol repository, I downloaded OrganiX13. I removed duplicate entries and split the dataset into training, validation, and testing sets in a 8:1:1 ratio.