Does Neural Style Transfer needs huge RAM?

So after having a thread that is solved by @Suleman_Navalur , i started to see that it takes a lot of RAM. I even use Google Colab which gives free GPU, but it even says that the session crash. Does it really need a huge RAM just to train?

If you are using the same notebook as Coursera platform, and training a single image for style transfer, I do not think it’s memory hungry. Typical case that needs large RAM is a larger batch size, but this is not the case. I’m using 64GB machine with AMD GPU locally, and checked resource usage. This assignment uses GPU fully, but the usage of CPU memory is less than 1GB. Is it GPU memory that you are talking about ?

i started to see that it takes a lot of RAM

What is your observation that concludes a memory shortage ?

Yes. I actually have some wrong lines of code putted in my code that caused that specific error. Thanks for clarifying!