diff --git a/content/blog/cuda_net.md b/content/blog/cuda_net.md
index 1b70993..a679c10 100644
--- a/content/blog/cuda_net.md
+++ b/content/blog/cuda_net.md
@@ -15,3 +15,7 @@ Quickly discovering that without writing custom kernels, you can't really progre
 - padding kernel 
 - column major / row major headache
 - removing cuBLAS -> just row major representation
+- naive conv2d
+- learning 3D memory representation
+- optimizing conv2d
+- softmax sum reduce
\ No newline at end of file