In many tensorflow tutorials (example) "towers" are mentioned without a definition. What is meant by that?
Asked
Active
Viewed 5,494 times
2 Answers
8
According to tensorflow documentation about CNN,
The first abstraction we require is a function for computing inference and gradients for a single model replica. In the code we term this abstraction a "tower".
To get the relevant context and more, check this.
Hima Varsha
- 2,316
- 14
- 34
-
1The linked page does not mention "tower" anymore :/ – aviator Apr 07 '20 at 15:17
-
To keep it simple, a "Tower" is a copy of your model. In a multi-GPU case, if you have 8 GPUs, then you will get 8 Towers. – Begoodpy Jul 26 '20 at 13:13