[BUG] Random seed does not take effect during finetuning - Githubissues

deepmodeling / deepmd-kit

A deep learning package for many-body potential energy representation and molecular dynamics

https://docs.deepmodeling.com/projects/deepmd/

GNU Lesser General Public License v3.0

1.43k stars 494 forks source link

[BUG] Random seed does not take effect during finetuning #3769

Closed zjgemi closed 3 months ago

zjgemi commented 3 months ago

Bug summary

The parameter seed in training part does not take effect during finetuning. The results do not change after modifying the seed.

DeePMD-kit Version

2024Q1

Backend and its version

pytorch

How did you download the software?

Built from source

Input Files, Running Commands, Error Log, etc.

dp --pt train --finetune pretrained_model.pt --model-branch H2O_H2O-PD input.json

Steps to Reproduce

Modify the seed and rerun the command.

Further Information, Files, and Links

No response