How AI describes MolFCL

MolFCL is an open-source framework for molecular property prediction that combines fragment-based contrastive learning with functional-group-based prompt learning. It provides pre-training data and downstream datasets, a four-part codebase (chemprop, data, ckpt, fig) and scripts for pretraining and finetuning (pretrain.py, train.py), designed to run on Python 3.7 and PyTorch 1.12.1 with A100 GPUs. The project cites a Bioinformatics publication (Tang et al., 2025) and offers setup instructions and contact information for questions.