Pytorch Quick Tip: Reproducible Results and Deterministic Behavior

In this video I show you how to set the seeds and what you need to do in order to obtain deterministic behavior from Pytorch (note this is not cross platforms). This is very useful when comparing changes you've made (to see that it actually does a difference) and debugging and so on!

❤️ Support the channel ❤️

Paid Courses I recommend for learning (affiliate links, no extra cost for you):

✨ Free Resources that are great:

💻 My Deep Learning Setup and Recording Setup:

GitHub Repository:

✅ One-Time Donations:

▶️ You Can Connect with me on:

[fuat]

That hitchiker’s guide to galaxy referance was great 😀

[mehdiorouji]

where in the training loop should we set these random seeds? before each eapoch, or before each batch, or even before the training loop?

[deepshankarjha]

this is something you will not get even in most of the paid courses. thanks a lot alladin

[donkkey]

Really enjoy your courses. I must say yours are better than some college courses. You should higher your membership rate so you can have more time for making videos. Thanks!

[tawheedrony]

Hey I am just working on a competition in Kaggle. But facing a ton of difficulties in optimizing hyper parameters and fine tuning the model. Can you please suggest something to do? Or planning on making a video? Thanks as always. 😌

[dataaholic]

Looks like this video is for me 😂 . Thanks for this video .

[bnnibell_]

what does seed=42 means? some did is it different?

[ximingdong]

thanks brother, also we need a torch.cuda.backends.enable = False.