[PYTHON] Tensorflow-GPU seems to be together if it is TF2.0 or later?

Is this also written using html or something? I'm ignorant so I'm not sure

If you're running Autokeras, maybe you're not using a GPU? I thought it wasn't really used.

Until now, I thought that GPUs would be used without permission. But The CPU performance is not so low, so I didn't think it was too late. No, it's a story compared to what.

So there's Tensorflow-gpu! I thought I would do my best to put it in It seems that the versions of CUDA and CuDNN must also be matched. It worked, but apparently Autokeras demands TF2.3 or something with a margin. I thought something was wrong.

So Autokeras didn't work after all, but if you look closely It seems that they are together after TF2.0. Certainly TF 1.15 and below are individual I saw something like that, but I didn't understand what it meant and ignored it. Stupid.

I saw that tensorflow is overwritten when tensorflow-gpu is inserted, That was the story at that time.

I haven't tried it yet, but for the time being, I'll try to build an environment so that TF2.3 works. Also, check if it is the version of the article you use. I wasted a lot of time, but I learned a lot. Complete

Recommended Posts

Tensorflow-GPU seems to be together if it is TF2.0 or later?
If it is not easy to understand, it cannot be improved.
How to use any or all to check if it is in a dictionary (Hash)
python> Is it possible to make in-line comments?> It seems that it has to be on multiple lines
Did love happen to be born? Is it someone's plot?
I made an AI to judge whether it is alcohol or not!
If pip stops due to SSL, it is better to re-insert python itself
[For Numba beginners] It is better to set nopython = True to check if the speedup is realized (for version 0.50.0 or earlier)
Notes on what to do if "macOS 11 or later required!" Appears in Big Sur or pyarrow2.0.0 cannot be installed