When training my model normally with ML Compute enabled everything runs fine. However when disabling ML Compute I get a number of errors in different layers of different models.
Dropout Layer Error:
2021-03-08 09:52:49.148915: W tensorflow/core/framework/op_kernel.cc:1775] OP_REQUIRES failed at mlc_dropout_ops.cc:79 : Aborted: Compute: Operation received an exception: Compute: VerifyImpl: Cached MLCLayer is not a type of Dropout.
Sigmoid Layer Error:
tensorflow.python.framework.errors_impl.InvalidArgumentError: assertion failed: [predictions must be >= 0] [Condition x >= y did not hold element-wise:] [x (sequential/dense_3/Sigmoid:0) = ] [[nan][nan][nan]...] [y (Cast_8/x:0) = ] [0]
[[{{node assert_greater_equal/Assert/AssertGuard/else/_1/assert_greater_equal/Assert/AssertGuard/Assert}}]] [Op:__inference_train_function_89873]
As I said these models work fine with ML Compute enabled.
When training my model normally with ML Compute enabled everything runs fine. However when disabling ML Compute I get a number of errors in different layers of different models.
Dropout Layer Error:
Sigmoid Layer Error:
As I said these models work fine with ML Compute enabled.
I am running on a 2020 M1 Macbook Pro.