Implement output layer

test/layers/test_output.cu

@@ -0,0 +1,24 @@
+#include <cuda_runtime_api.h>
+#include <gtest/gtest.h>
+
+#include "output.cuh"
+
+TEST(OutputLayerTest, OutputForward) {
+    cudaError_t cudaStatus;
+
+    std::vector<float> input = {0.573f, 0.619f, 0.732f, 0.055f, 0.243f, 0.316f};
+    float*             d_input;
+    cudaStatus = cudaMalloc((void**)&d_input, sizeof(float) * 6);
+    EXPECT_EQ(cudaStatus, cudaSuccess);
+    cudaStatus = cudaMemcpy(
+        d_input, input.data(), sizeof(float) * 6, cudaMemcpyHostToDevice
+    );
+    EXPECT_EQ(cudaStatus, cudaSuccess);
+
+    CUDANet::Layers::Output outputLayer(6);
+    float* h_output = outputLayer.forward(d_input);
+
+    for (int i = 0; i < 6; ++i) {
+        EXPECT_EQ(input[i], h_output[i]);
+    }
+}