Deep Learning Hyperparameter Tuning in PyTorch | Making the Best Possible ML Model | Tutorial 2

[@GregHogg]

Take my courses at mlnow.ai/!

[@Pacificatorrr]

you're an absolute legend, thank you for this!

[@arsheyajain7055]

Amazing video!! Thank you 😊

[@GregHogg]

And thank you for the support :)

[@emiliodaza2902]

Thank you for your valuable contribution!

[@danieltunnermann]

Hi, nice video, thanks. I think you defined the dropout layer but didn’t apply it in the forward pass.

[@dharmiknaik1772]

Good content. Thank you Greg.

[@GregHogg]

Glad you enjoyed it!!

[@nickchern392]

Dude, u've saved me❤

[@GregHogg]

Woot! 🥰

[@dancode0775]

Hey Greg ,amazing Video !
Do you have content that covers database since SQL is part of a data scientist job ?

[@GregHogg]

Thank you! Yes I do, you can search my channel for SQL content.

[@slavavoiteshonok3648]

Hey, Greg, thanks a lot for this video
Could you tell please your jupyter notebook setup that looks so pretty?

[@GregHogg]

Thank you! I think I just googled something like dark mode for jupyter and followed some quick instructions.

[@rhythmdutta387]

Extremely informative video! Would you mind sharing your pc specs Greg?? Specifically the gpu and its vram..

[@Sharky-1507]

Rtx 4080 with the 13 gen i7 cpu, 32gb ram

[@stanley_george]

Why are you shutting down your kernel instead of simply changing the value and running the next blocks ?

[@software_solve]

Dear thank you for uploading informative videos. Can you share link or make video on Random forest implementation using pyTorch. Thanks

[@GregHogg]

Pytorch is for deep learning, random forest is a classical algorithm

[@nickchern392]

Greg, could you share your's PC setup?

[@GregHogg]

It's brand new, pretty much state of the art. Rtx 4080 with the 13 gen i7 cpu, 32gb ram

[@vitorpmh]

The gpu don't explode because the 28x28 images occupy low memory. I have a 1070ti and i was surprised as well. However if you try the imagenet 224x224x3 images if you have a high batch size such as 512 it explode.s

[@GregHogg]

Yeah, absolutely, the bigger it gets the more likely it will boom boom