How to ensure reproducibility of results

mit-han-lab / torchsparse

[MICRO'23, MLSys'22] TorchSparse: Efficient Training and Inference Framework for Sparse Convolution on GPUs.

https://torchsparse.mit.edu

MIT License

1.15k stars 131 forks source link

How to ensure reproducibility of results #284

Closed ybc-ybc closed 4 months ago

ybc-ybc commented 6 months ago

In version torchsparse v2.1, do fixed seeds guarantee reproducible results?

ys-2020 commented 6 months ago

Hi @ybc-ybc , we don't rely on the fixed seeds since all the benchmark data used in our paper are real-world data, which means that the inputs are not random.

ys-2020 commented 4 months ago

Close as completed due to inactivity. If you have further questions, feel free to reopen it.