diff --git a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
index ec74c1c2d54a..66362d5be4f1 100644
--- a/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
+++ b/mlir/lib/Dialect/SparseTensor/Pipelines/SparseTensorPipelines.cpp
@@ -32,7 +32,6 @@
 void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
                                           const SparsifierOptions &options) {
   // Rewrite named linalg ops into generic ops.
-
   pm.addNestedPass<func::FuncOp>(createLinalgGeneralizationPass());
 
   // Sparsification and bufferization mini-pipeline.
@@ -65,6 +64,8 @@ void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
     pm.addNestedPass<gpu::GPUModuleOp>(createConvertGpuOpsToNVVMOps());
   }
 
+  // Progressively lower to LLVM. Note that the convert-vector-to-llvm
+  // pass is repeated on purpose.
   // TODO(springerm): Add sparse support to the BufferDeallocation pass and add
   // it to this pipeline.
   pm.addNestedPass<func::FuncOp>(createConvertLinalgToLoopsPass());
@@ -80,10 +81,7 @@ void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
   pm.addNestedPass<func::FuncOp>(createConvertMathToLLVMPass());
   pm.addPass(createConvertMathToLibmPass());
   pm.addPass(createConvertComplexToLibmPass());
-
-  // Repeat convert-vector-to-llvm.
   pm.addPass(createConvertVectorToLLVMPass(options.lowerVectorToLLVMOptions()));
-
   pm.addPass(createConvertComplexToLLVMPass());
   pm.addPass(createConvertVectorToLLVMPass(options.lowerVectorToLLVMOptions()));
   pm.addPass(createConvertFuncToLLVMPass());
@@ -101,6 +99,7 @@ void mlir::sparse_tensor::buildSparsifier(OpPassManager &pm,
     pm.addPass(createGpuModuleToBinaryPass(gpuModuleToBinaryPassOptions));
   }
 
+  // Ensure all casts are realized.
   pm.addPass(createReconcileUnrealizedCastsPass());
 }