README.md

Code to obtain the $\mathrm{SLATM}_d$ results

run_slatm.py — the main script that computes the representation (if not in repr/), optimizes the hyperparameters (if not in hypers.py), and computes the prediction errors (if not in results/)
environment.yml — conda environment file
run.bash — wrapper script to recompute all the results
hypers.py — models hyperparameters
manh.c, makefile — optional C module for faster computation of Laplacian kernel
learning.py, reaction_reps.py — functions to perform KRR and compute $\mathrm{SLATM}_d$, respectively

repr/ — computed representations (not in the repo because of the size)
results/ — saved MAE/RMSD
by_reaction/ — predictions for each test reaction: reaction index (in the dataset), prediction