A question about the init-model mode

[TerryHangLyu]

Dear developers and users,
I have a question when browzing the Zhihu introduction of DeepMD-kit. That's is how can we know that the prossibility for the old data set and new data set is 0.1 and 0.9 respectively from the keywords.
From this part ("auto_prob_style": "prob_sys_size; 0:3:9; 3:5:1"), I can not see how the prossibility for the old data set and new data set is defined.

[njzjz]

That's is how can we know that the prossibility for the old data set and new data set is 0.1 and 0.9 respectively from the keywords.

The probability is 0.9 and 0.1, respectively, instead of 0.1 and 0.9.

[TerryHangLyu]

Thank you!

[likefallwind]

You can check this：https://docs.deepmodeling.org/projects/deepmd/en/master/train/training-advanced.html

[TerryHangLyu]

Thank you!

[likefallwind]

We also add some explaination in Chinese：https://zhuanlan.zhihu.com/p/348671837?