---
dataset_info:
  features:
  - name: SMILE
    dtype: string
  - name: SELFIE
    dtype: string
  splits:
  - name: train
    num_bytes: 351931924.06659317
    num_examples: 1273077
  - name: val
    num_bytes: 21949894.491152223
    num_examples: 79564
  - name: test
    num_bytes: 65951655.37470361
    num_examples: 238694
  download_size: 148629975
  dataset_size: 439833473.932449
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: val
    path: data/val-*
  - split: test
    path: data/test-*
---

# Dataset Card for Guacamol

Dataset from the [Guacamol](https://github.com/BenevolentAI/guacamol) benchmark ([paper](https://arxiv.org/abs/1811.09621)).

Dataset contains two columns, SMILE and SELFIE. Splits are identical to original splits, however, any SMILE that could not be converted to a SELFIE was dropped. Likewise, any SELFIE in the val/test splits that contained a token not found in the train split was dropped.

Can be used with [this tokenizer](https://huggingface.co/haydn-jones/GuacamolSELFIETokenizer).