[FEA] Mixed Precision Support for Merlin Models TensorFlow

Open bschifferer opened this issue 3 years ago • 1 comments

🚀 Feature request

Oct 13 '22 16:10 bschifferer

@bschifferer did you get a chance to test these two lines and see some speed up?

policy = mixed_precision.Policy('mixed_float16')
mixed_precision.set_global_policy(policy)

Oct 24 '22 15:10 rnyak