Tensorflow U-Net using way too much memory
I'm training a U-Net CNN on a custom dataset, but whenever I train the model, I get an error saying that the memory resources have been exhausted. This does not make sense, though, as my batch size is only 8 (512 x 512 png images).
I am working on VS code with a CUDA-enabled GPU.
I have already tried some popular fixes such as:
tf.compat.v1.reset_default_graph()
config = tf.compat.v1.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.compat.v1.InteractiveSession(config=config)
but nothing has worked
Comments
Post a Comment