Are there any side-effects of initing tf.compat.v1.Session with tensorflow 2?

TensorFlow
,
1

I am using this code in Python with tensorflow=2.11 to set allow_growth = True:

            config = tf.compat.v1.ConfigProto()
            config.gpu_options.allow_growth = True
            session = tf.compat.v1.Session(config=config)

My question is if there are any other (side-)effects on the tensorflow behavior except for setting the memory allocation to be dynamic.

2

Hi @clime, Generally it is not preferable to use the compat.v1 module with 2.x as it might not guarantee to work as expected because the module was deprecated in 2.x. Thank You.

3

Thanks, though, I am not asking about anything “generally”. I am asking about concrete side effects of using that code with tensorflow 2 (or tensorflow 2.11 if that helps).