Table 5 Performance of IUPAC to SMILES models trained on increasing training data
From: STOUT V2.0: SMILES to IUPAC name conversion using transformer models
Training data size | Valid SMILES | Identical predictions | Tanimoto 1.0 | Average tanimoto |
|---|---|---|---|---|
100 million | 99.80% | 96.97% | 99.12% | 0.99 |
 ~ 1 billion | 99.60% | 96.23% | 99.06% | 0.99 |