deepmd-train-dpa3

Train DeePMD-kit models with the DPA3 descriptor using dp --pt train.

124|25|Updated Feb 20, 2026

Search Tags:#gpu #training #machine-learning #deepmd-kit #dpa3 #materials-simulation #line-graph-series

Ready to run

npx skills add https://github.com/jinzhezenggroup/computational-chemistry-agent-skills --skill deepmd-train-dpa3

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: deepmd-train-dpa3
Source: https://github.com/jinzhezenggroup/computational-chemistry-agent-skills/tree/main/machine-learning-potentials/deepmd-train-dpa3
Command: npx skills add https://github.com/jinzhezenggroup/computational-chemistry-agent-skills --skill deepmd-train-dpa3

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Train a DeePMD-kit model using the DPA3 descriptor to achieve high-accuracy neural potentials for large atomic systems across diverse chemical and material environments.

Core Features & Use Cases

Supports training with the DPA3 descriptor on Line Graph Series (LiGS) for large atomic models.
Handles multi-element datasets, configurable neighbor selection, and PyTorch-backed training.
Use cases include developing transferable potentials for materials simulations and exploratory research.

Quick Start

Train a DPA3-based DeePMD-kit model by providing a prepared input.json to the dp --pt train command.

Frequently Asked Questions about deepmd-train-dpa3

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I train a DeePMD-kit model with the DPA3 descriptor using PyTorch?▼

To train a DeePMD-kit model with the DPA3 descriptor, prepare an input.json file and execute the dp --pt train command with a working PyTorch backend to build high-accuracy neural potentials for materials simulations.

What is the DPA3 descriptor used for in machine-learning materials simulations?▼

The DPA3 descriptor is used in machine-learning materials simulations to train high-accuracy neural potentials, supporting large atomic systems and diverse chemical environments across multi-element datasets.

Can I use the DPA3 descriptor for multi-element datasets in DeePMD-kit?▼

Yes, training with the DPA3 descriptor in DeePMD-kit handles multi-element datasets and configurable neighbor selection, enabling the development of transferable potentials for complex materials simulations.

What's the best way to build transferable neural potentials for large atomic systems?▼

The best way to build transferable neural potentials for large atomic systems is training with the DPA3 descriptor on Line Graph Series (LiGS) within the DeePMD-kit framework using a PyTorch backend.

Do I need a specific backend to run DPA3 training workflows in DeePMD-kit?▼

Yes, DPA3 training workflows in DeePMD-kit require a working installation of the software with the PyTorch backend to execute the dp --pt train command and monitor the training process.