---
dataset_info:
  features:
  - name: formula
    dtype: string
  - name: smiles
    dtype: string
  splits:
  - name: test
    num_bytes: 535072
    num_examples: 5000
  download_size: 262346
  dataset_size: 535072
configs:
- config_name: default
  data_files:
  - split: test
    path: data/test-*
license: cc-by-nc-sa-4.0
language:
- en
tags:
- chemistry
- coconutdb
- SMILES
- chemteb
pretty_name: CoconutDB SMILES to Formula Bitext Mining
size_categories:
- 1K<n<10K
---
# CoconutDB SMILES to Formula Bitext Mining
This dataset consists of two lists: one containing both isomeric and canonical SMILES strings, and the other containing the corresponding molecular formulas of chemical entities, sourced from [CoconutDB](https://coconut.naturalproducts.net/). The primary task is to identify matching pairs between the SMILES strings and their molecular formulas. Each SMILES string from the first list should be accurately aligned with its corresponding molecular formula from the second list. This dataset is a subset of [BASF-We-Create-Chemistry/CoconutSmiles2NameBitextMining](https://huggingface.co/datasets/BASF-We-Create-Chemistry/CoconutSmiles2NameBitextMining) with 5000 samples.