why predict result is different in same model and environment?

[Helloworld0214]

Thank for your excellent work. I use PICK to train invoices, but why the predict result is different in same model and environment? Looking forward to your reply.

[Helloworld0214]

I got the answer. The test process need add random seed. @wenwenyu

[jorgerodriguezsj]

@Helloworld0214 Where do you set the random seed?

[Helloworld0214]

@Helloworld0214 Where do you set the random seed?

I set it before in test model loaded.

[IrvingHerreraLuna]

I have the same problem, I trained with GPU on an EC2 instance, now I am running it on CPU and the test results change, Can I be more explicit in the use of the seed? This is during the test? Even inserting the seed the result is different

[ImmineoII]

Hi, can you post the set_seed() function? @Helloworld0214