Published checkpoints

Model weights use Git LFS. Run git lfs pull after cloning.

Ir checkpoints.pth: Ir pretrained model; selected CV fold 5; fixed 251-record test set.
Co transfer checkpoints.pth: Co transfer model initialized from Ir; selected fold 1, stage 2; fixed 14-record test set.

Both models use the same three-graph architecture and latent H/S head; select the matching xTB directory and catalyst when predicting.

python scripts/predict.py --model "models/Ir checkpoints.pth" --input "data/ir_holdout/Database Ir Test.csv" --output outputs/ir_predictions.csv --cpu
python scripts/predict.py --model "models/Co transfer checkpoints.pth" --input config/co_reference/test.csv --xtb_root data/xtb_Co_features --default_catalyst Co --output outputs/co_predictions.csv --cpu

Run commands from the repository root.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support