Hi team,
I have been trying to port code that uses tf.RaggedTensor on my M1 machine, but it generates segmentation fault errors.
I managed to replicate and isolate the issue with the following code:
import tensorflow as tf
x = tf.ragged.boolean_mask([[0.]],[[False]])
tf.reduce_mean(x,axis=1)
Running these lines on my macbook produces a segmentation fault.
Expected output, tested on a Linux machine with Tensorflow 2.6.0
<tf.Tensor: shape=(1,), dtype=float32, numpy=array([nan], dtype=float32)>
My configuration
- Python 3.9.6
- tensorflow-metal 0.1.2
- tensorflow-macos 2.5.0
- Big Sur 11.5.1
- MacBook Pro (13-inch, M1, 2020)