That's a common feeling because setting up tech is daunting. Tools like Tensorflow are developed at companies like Google for use by teams of engineers...or individuals with access to support teams. Companies like Google don't design products like Tensorflow to hide all the necessary complexity and don't document the products for less technical users.
When a product like Tensorflow is released by Google, many early adopters are working in environments with institutional knowledge about sophisticated tools. For them, Tensorflow is easy to use relative to what they were using before but it's not easy to use like an iPhone or other consumer technology.
So what might this mean for you? Well if setting it up is daunting you will learn a lot by doing it even if you never really get it set up quite right or abandon it as too much bother or become an expert. Or to put it another way, when technology X requires technology Y which requires technology Z, maybe I wind up learning more about Y and Z than X and get back to X in a year or two once I have the background and if I don't come back then it is sometimes that X wasn't really as interesting as I thought but going through the process makes picking up technology W easier.
So my random advice from the internet is stick with it because it is daunting and see where it goes. One of my favorite essays is by Peter Norvig who is a Principal Engineer at Google: http://norvig.com/21-days.html
Python version 3 works well for me. Hardware wise as of right now a nvidia cuda card compute level 3.0 or greater is needed for running on gpu. That means basically a Ge-force 710 or better, also tensorflow runs on a cpu (for now) but that does affect performance.
When a product like Tensorflow is released by Google, many early adopters are working in environments with institutional knowledge about sophisticated tools. For them, Tensorflow is easy to use relative to what they were using before but it's not easy to use like an iPhone or other consumer technology.
So what might this mean for you? Well if setting it up is daunting you will learn a lot by doing it even if you never really get it set up quite right or abandon it as too much bother or become an expert. Or to put it another way, when technology X requires technology Y which requires technology Z, maybe I wind up learning more about Y and Z than X and get back to X in a year or two once I have the background and if I don't come back then it is sometimes that X wasn't really as interesting as I thought but going through the process makes picking up technology W easier.
So my random advice from the internet is stick with it because it is daunting and see where it goes. One of my favorite essays is by Peter Norvig who is a Principal Engineer at Google: http://norvig.com/21-days.html
Good luck.