提交 5388b228 编写于 作者: P Patrick Nguyen 提交者: TensorFlower Gardener

Require min/max operands to be on the CPU.

Change: 150028654
上级 3971c27b
......@@ -164,6 +164,8 @@ TF_CALL_double(REGISTER_CPU_KERNEL);
#define REGISTER_GPU_KERNEL(T) \
REGISTER_KERNEL_BUILDER(Name("QuantizeAndDequantizeV2") \
.Device(DEVICE_GPU) \
.HostMemory("input_max") \
.HostMemory("input_min") \
.TypeConstraint<T>("T"), \
QuantizeAndDequantizeV2Op<GPUDevice, T>); \
REGISTER_KERNEL_BUILDER( \
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册