w86763777 / pytorch-ddpm

Unofficial PyTorch implementation of Denoising Diffusion Probabilistic Models
Do What The F*ck You Want To Public License
506 stars 62 forks source link

Request to Add Random Seed Setting for Reproducibility #21

Open yanliang3612 opened 10 months ago

yanliang3612 commented 10 months ago

I hope this message finds you well. I am writing to bring to your attention an aspect of the codebase that could be enhanced for better reproducibility and consistency in results.

I noticed that the current implementation in the [specific file or section of the project] does not specify a random seed for operations that involve randomness. This absence can lead to variations in outcomes each time the code is executed, making it challenging to reproduce results or compare changes effectively.

For research and development purposes, having a consistent and reproducible environment is crucial. Setting a random seed ensures that the random number generation is consistent across different runs, which is essential for debugging and comparing algorithm performance.

It would be greatly beneficial for users of this repository if a feature could be added to set a fixed random seed, either as a default setting or through a user-specified option. This change would enhance the usability and reliability of the project.

Thank you for considering this request. Your work on this project is greatly appreciated, and I look forward to any updates on this matter.

Best regards, Divin