Abstract
An algorithm for the black-box generation of high-quality system-specific machinelearning interatomic potentials (MLIPs) for gas phase reactions in the electronic ground state is presented. It relies on the self-consistent fine-tuning of a MLIP foundation model, where the message-passing atomic cluster expansion (MACE) is taken as an example, based on data collected from biased samplings along the reaction coordinate of interest with the Caracal program. The reaction is first sampled with the foundation model at different temperatures, and then the sampling is repeated with the fine-tuned model until the errors of selected energies and gradients with respect to the high-level reference method falls below a chosen threshold. In this paper, this method is benchmarked systematically by alternating the amount of collected training data and the MACE MLIP architecture, such that an optimal compromise between speed and accuracy can be obtained. To make a direct comparison of reaction rate constants between MLIP and the reference method possible, two gas phase reactions with analytical potential energy surfaces from the literature have been chosen: The internal proton transfer in malonaldehyde and the proton exchange between methane and an