An independent study has demonstrated that because of the streamlined syntax of BigSMILES, polymer ML workflows based on BigSMILES consistently required shorter training times compared to those based on SMILES, particularly in large language model scenarios. The authors conclude that as datasets for polymer ML training grow, using BigSMILES as a representation could significantly accelerate the construction of polymer ML pipelines, whether in forward screening paradigms or inverse design paradigms.
top of page
bottom of page
Comments