Am just starting the fast.ai course (Part 1 2020).
Am from India & am concerned about the monthly costs involved in running the AWS EC2 instance, required to finish the course.
So, can anyone please help me in this regard and reply back (at convenience) with rough estimates of the below, so that I could plan for my monthly spend on AWS EC2 :
How many hours of running AWS EC2 instance is required (on a daily basis, on monthly basis, overall) to finish the fast.ai course (both parts)?
What might be the at-most monthly bill of AWS EC2, if we try to finish the course (both parts) as per the course schedule?
How many months it might take to complete the course (both parts) as per the prescribed schedule.
Hi mate, it took me about 2 months to get through Part 1, but then again I have a full time job. So I can only do about 3-4 hours a day / 5 days a week. So your mileage may vary according to how much free time you have… the main reason why it took me so long is because I kept doing research as I learned stuff from the chapters and went back over certain parts. So you could probably do it faster, especially if you have a mathematical background.
As far as costs go. I spend about $8 USD a month on Paperspace Gradient. But some people on the forums use Google Collab for free, both work, but make sure to follow the documentation on how to set them up for your first lesson. In my opinion AWS is overkill for learning, especially if you’re just starting out on Deep Learning.
I agree with @jquintanilla4, aws is not necessary if you are just learning.
if you really want to do aws, another option would be to do all your training on google colab, which is free, and host the models on aws. you might be able to get away with the free tier of aws, so your cost could be 0.