Fastai GPU Platform/Cloud computing setup

I have been trying different machines and platforms, fastai, fastai2, in colab, ubuntu in virutal box in win10 and also Conda in Win10 of my notebook with 1050Ti . I also tried gradient.

The following are the lessons that I learned are going along with the mainstream, especially for a beginner, like me.

  1. Looking for a different platform, comparing the pricing of different usage, different degrees of setting it up, etc … can take you a lot of time.
  2. Use ubuntu in Vbox seems to be a good option for me. My Window 10 and office is useful with excel in the later lessons. I may try Jeremy’s suggestion of ubuntu in Win later, but my HD has only 20 percent left.
  3. Use Google Cloud, most people using it and quite likely have longer support, got $300 quota-free.
  4. Following what the developer is using can be a safe path, but that is against the keep trying and innovative principles. Try to figure out your mind, which way you want to go, swaying back and forth get you to nowhere.
  5. I am setting up my GCP and hoping and waiting for the approval of my instance, it will work fine.
  6. For browsing through the notebook occasionally, I think collab can still be good free resources and attached to the google drive for storage.