Replication Data for: How to Train a Shallow Ensemble

Le résumé fourni par la source

This repository accompanies our paper on "How to Train a Shallow Ensemble". It includes workflows and input files needed to reproduce the results. Most of the experiments are contained in the directory "0_convergence_nll_training". There is a separate directory for the experiments performed for the first revision of the publication and for the repetition of an experiment with an extended version of the dataset. Each directory contains a pyproject.toml and a uv.lock in order to recreate the python environment used for the experiments.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.