diff --git a/docs/radix-memory.md b/docs/radix-memory.md
index 9252866..1bdd193 100644
--- a/docs/radix-memory.md
+++ b/docs/radix-memory.md
@@ -35,7 +35,7 @@ New benchmark suite is added, which measures the latency of the following operat
 
 - Memory read / write to random addresses
 - Memory read / write to contiguous address
-- Memory write to sparse memory addresse
+- Memory write to sparse memory addresses
 - Memory write to dense memory addresses
 - Merkle proof generation
 - Merkle root calculation
@@ -112,4 +112,4 @@ Usually, sparse region would utilize smaller branching factor for memory optimiz
 - use larger branching factors at the upper address level to reduce the trie traversal depth
 - use smaller branching factors at the lower address level to reduce computation for each node.
 
-In addition, we can apply pgo as mentioned above. To apply pgo to asterisc builds, we can run asterisc with cpu pprof enabled, and ship asterisc with `default.pgo` in the build path. This way, whenever the user builds Asterisc, pgo will be enabled by default, leading to addition 5+% improvement in speed.
\ No newline at end of file
+In addition, we can apply pgo as mentioned above. To apply pgo to asterisc builds, we can run asterisc with cpu pprof enabled, and ship asterisc with `default.pgo` in the build path. This way, whenever the user builds Asterisc, pgo will be enabled by default, leading to addition 5+% improvement in speed.