Re-evaluating Benchmarks for MLIPs

This repository contains files related to the data generation, training, and evaluation for MLIP models. The repository is organized as follows:

Directory Structure

• Dataset Generation/
• MLIP Training/
  • Allegro MLIP Models
  • DeePMD MLIP Models
• MLIP Evaluation/
  • Allegro MLIP Models
  • DeePMD MLIP Models

Go to each subfolder for a detailed description.

Dataset Generation

The generate-datasets/ directory contains scripts to generate datasets for MLIP model training.

MLIP Training

Allegro MLIP Models

The training-allegro/ directory provides datasets and sample configuration files for the training process of Allegro MLIP models.

DeePMD MLIP Models

The training-deepmd/ directory provides datasets and sample configuration files for the training process of DeePMD MLIP models.

MLIP Evaluation

Allegro MLIP Models

The evaluate-allegro/ directory contains scripts to perform FES estimation with Allegro.

DeePMD MLIP Models

The evaluate-deepmd/ directory contains scripts to perform FES estimation with DeePMD.