I am reading the course1 notebook.
Although I have nvidia GPU laptop it won’t support cuDNN.
In this case whether installing pytorch with cuda 8.0 will be of any help to work on cpu only mode.
Or installing pytorch without cuda alone is enough to work.
I understand we need to work on GPU enabled server to complete the lab. But I like do some basic homework before switching to cloud instance to save on cost.
Please provide your inputs.
You can absolutely learn pytorch and experiment with various technologies used in this course on CPU only.
Hi Prabhu,
Installing cuda will be of no help, if you are planning to work on CPU.
Cuda drivers are specifically designed libraries for shifting PyTorch variables to GPU for faster processing.
Check this Installing cuda on cpu on stackoverflow.
Hope this helps.
Hi Radek & Poonam,
I am installing pytorch without cuda. Thanks.
You could also refer to my medium https://medium.com/towards-data-science/my-first-step-for-deep-learning-adventure-with-udacity-and-coursera-ee135042ac1e and https://medium.com/towards-data-science/tensorflow-jupyter-notebook-nvidia-diy-setup-473acfe5b0e for some additional software especially the monitoring, letsencrypt and vnc part.
I am also having laptop with GTX 1050. will it be sufficient? I am planning to install Ubuntu on windows Hyper-V. after that what all do I need to setup besides anaconda as in intro workshop, for completing lesson 1?