[minimaxir]

I was initially skeptical of Colaboratory notebooks when it was first released, but the value of a free GPU (and Python 3 support eventually added) makes it worthwhile to use. I recently wrote a tutorial on training a text-generating neural network based on Colaboratory here: http://minimaxir.com/2018/05/text-neural-networks/ (and I have trained many-hours-worth of models with it!)

The only problem I have with the Colaboratory Notebooks is that file I/O is a pain in the butt (and requires Chrome).

[mtyka]

fwiw we have an example notebook for I/O:

https://tools.google.com/seedbank/seed/5646239437684736

Transfer from Google Drive is particularly fast (many GB is no problem) since it's google datacenter to datacenter.

[ma2rten]

If you don't mind me asking: It seems like you can just run arbitrary stuff in a cloud instance for free using colab including GPUs. This doesn't sound a like a sustainable offering.

This sounds a lot like the old Google, which offered a lot of stuff for free but had to shut it down. Do you have any thoughts on that? Are there plans for a paid offering?

[jorgemf]

It sounds to me they want people to learn about deep learning and tensorflow. In the long term there will be more professionals and the salaries will go down. That is how they will make profit from this type of thing.

[mkl]

Unrelated, but since you're here: The animated background on that site seems completely unnecessary and uses a lot of resources - it makes my laptop's fans go crazy.

[dotancohen]

That is why Colaboratory offers the free GPU!