diff --git a/include/layers/conv2d.cuh b/include/layers/conv2d.cuh
index e30321e..3ec73ca 100644
--- a/include/layers/conv2d.cuh
+++ b/include/layers/conv2d.cuh
@@ -25,6 +25,9 @@ class Conv2d {
     int outputSize;
 
     void forward(const float* d_input, float* d_output);
+    void setKernels(const std::vector<float>& kernels_input);
+
+    void host_conv(const float* input, float* output);
 
   private:
     // Inputs
@@ -49,10 +52,6 @@ class Conv2d {
 
     void initializeKernels();
     void toCuda();
-
-    void setKernels(const std::vector<float>& kernels_input);
-
-    void host_conv(const float* input, float* output);
 };
 
 }  // namespace Layers
diff --git a/src/layers/conv2d.cu b/src/layers/conv2d.cu
index c31bc4a..8277c20 100644
--- a/src/layers/conv2d.cu
+++ b/src/layers/conv2d.cu
@@ -1,4 +1,5 @@
 #include <string>
+#include <iostream>
 
 #include "activations.cuh"
 #include "conv2d.cuh"
@@ -100,21 +101,25 @@ void Layers::Conv2d::host_conv(const float* input, float* output) {
                 
                 float sum = 0.0f;
 
+                // std::cout << "f: " << f << ", i: " << i << ", j: " << j << std::endl;
+
                 // Iterate over kernel and input matrix
                 for (int k = 0; k < kernelSize; k++) {
                     for (int l = 0; l < kernelSize; l++) {
                         for (int c = 0; c < inputChannels; c++) {
-
-                            // For now stride = 1
                             
                             int kernelIndex = k * (kernelSize * inputChannels * numFilters) + l * (inputChannels * numFilters) + c * (numFilters) + f;
-                            int inputIndex  = (i * stride + k) * (inputSize * inputChannels) + (j + stride + l) * (inputChannels) + c;
+                            int inputIndex  = (i * stride + k) * (inputSize * inputChannels) + (j * stride + l) * (inputChannels) + c;
+
+                            // std::cout << "kernelIndex: " << kernelIndex << ", kernel value: " << kernels[kernelIndex] << ", inputIndex: " << inputIndex << ", input value: " << input[inputIndex] << std::endl;
 
                             sum += kernels[kernelIndex] * input[inputIndex];
                         }                      
                     }
                 }
 
+                // std::cout << "sum: " << sum << std::endl;
+
                 output[i * (outputSize * numFilters) + j * (numFilters) + f] = sum;
             }
         }
diff --git a/test/layers/test_conv2d.cu b/test/layers/test_conv2d.cu
index 9e620d2..e4b675a 100644
--- a/test/layers/test_conv2d.cu
+++ b/test/layers/test_conv2d.cu
@@ -5,11 +5,11 @@
 
 #include "conv2d.cuh"
 
-TEST(Conv2dTest, ValidPadding) {
+TEST(Conv2dTest, SimpleExample) {
 
-    int inputSize = 3;
+    int inputSize = 4;
     int inputChannels = 1;
-    int kernelSize = 3;
+    int kernelSize = 2;
     int stride = 1;
     std::string padding = "VALID";
     int numFilters = 1;
@@ -28,8 +28,31 @@ TEST(Conv2dTest, ValidPadding) {
     int outputSize = (inputSize - kernelSize) / stride + 1;
     EXPECT_EQ(outputSize, conv2d.outputSize);
 
-    std::vector<float> input(inputSize * inputSize * inputChannels);
+    std::vector<float> input = {
+        1.0f, 2.0f, 3.0f, 4.0f,
+        5.0f, 6.0f, 7.0f, 8.0f,
+        9.0f, 10.0f, 11.0f, 12.0f,
+        13.0f, 14.0f, 15.0f, 16.0f
+    };
+    std::vector<float> kernels = {
+        1.0f, 2.0f, 3.0f, 4.0f,
+    };
+
+    conv2d.setKernels(kernels);
+
+    
     std::vector<float> output(outputSize * outputSize * numFilters);
-    std::vector<float> kernels(kernelSize * kernelSize * inputChannels * numFilters);
+
+    conv2d.host_conv(input.data(), output.data());
+
+    std::vector<float> expected = {
+        44.0f, 54.0f, 64.0f,
+        84.0f, 94.0f, 104.0f,
+        124.0f, 134.0f, 144.0f
+    };  
+
+    for (int i = 0; i < output.size(); ++i) {
+        EXPECT_FLOAT_EQ(expected[i], output[i]);
+    }
 
 }