This directory contains a simple reaction yield prediction demo using PyTorch and RXNFP fingerprints.

conda env create -f environment.yml
conda activate helsinki-pre-assignment

• yield_predict.py: Model and training loop
• rxn_data_processing_utils.py, rxn_logger.py: Data utilities and logging
• test_organic_reactions.csv: Example dataset

python yield_predict.py

Notes

• The script downloads/loads RXNFP model weights on first run.
• Training artifacts are saved to yield_predictor.pth every 10 epochs.
• For custom data, adapt column names to those used in yield_predict.py.