ConvolutionDepthwise ConvolutionDepthwise Inputs: input : float32|int8(NC4HW4) Outputs: output : float32 Backend: CPU Metal Vulkan OpenCL Caffe op: C...

Scale Scale Inputs: input : float32(NC4HW4|NHWC) Outputs: output : float32 Backend: CPU Metal Vulkan OpenCL Caffe op Scale BatchNorm CuDNNBatchNorm ...

QuantizedMaxPool QuantizedMaxPool Inputs: intput : uint8(NHWC) Outputs: output : uint8 Backend: CPU Metal Vulkan OpenCL Tensorflow Lite op: Quantized...

tfQuantizedConv2D tfQuantizedConv2D Inputs: intput : uint8(NC4HW4) Outputs: output : uint8 Backend: CPU Metal Vulkan OpenCL Tensorflow Lite op: Quant...

Const Const Inputs: 无 Outputs: output : float32|int32|int8 Backend: CPU Metal Vulkan OpenCL Tensorflow op: Const

Unpack Unpack Inputs: intput : float32(NHWC) Outputs: output0 : float32 outputN : float32 Backend: CPU Metal Vulkan OpenCL Tensorflow op: Unpack

Deconvolution Deconvolution Inputs: input : float32(NC4HW4) Outputs: output : float32 Backend: CPU Metal Vulkan OpenCL Caffe op: Deconvolution Tenso...

QuantizedLogistic QuantizedLogistic Inputs: intput : uint8(NHWC) Outputs: output : uint8 Backend: CPU Metal Vulkan OpenCL Tensorflow Lite op: Quantiz...

Selu Selu Inputs: input : float32(NHWC|NC4HW4) Outputs: output : float32 Backend: CPU Metal Vulkan OpenCL Tensorflow op: Selu

Supported optimizers Supported optimizers All standard Keras optimizers are supported, but importing custom TensorFlow optimizers won’t work: SGD RMSprop Adagrad Adadelta Ada...