train

Description:
Performs automatic training for NEP.

Input Parameters

Important

  • You should use NepTrain init <type> to generate job.yaml, and then use NepTrain train job.yaml to start the training task.

  • After the program runs, a restart.yaml file will be generated. To continue training, you can use NepTrain train restart.yaml.

Usage:

NepTrain train <config_path>

Options:

  • <config_path> Path to configuration file, such as job.yaml.

Output

During training, intermediate data and the resulting potential are written to the working directory. A restart.yaml file is created after each iteration, allowing you to resume the workflow with NepTrain train restart.yaml.

Example

Initialization

Run NepTrain init slurm on the command line to generate job.yaml, which contains all modifiable control parameters of the workflow.
The command produces a job.yaml file showing default execution parameters and their explanations. Set the parameters line by line on the first run, and later reuse the file as a template for future training.

Tip

If you copy a previously modified job.yaml, you can place it in the working directory and run NepTrain init slurm again to synchronize any newly added parameters during version updates.

Start Training

Run the following command on the login node:

NepTrain train job.yaml

To run in the background:

nohup NepTrain train job.yaml &