From: Rasmus Munk Larsen Date: Tue, 25 Jun 2019 21:59:40 +0000 (-0700) Subject: Fix typo in Toy tutorial documentation X-Git-Tag: llvmorg-11-init~1466^2~1333 X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=1df5f3159e538fdc12c451f09f3f879ca0200d66;p=platform%2Fupstream%2Fllvm.git Fix typo in Toy tutorial documentation PiperOrigin-RevId: 255054639 --- diff --git a/mlir/g3doc/Tutorials/Toy/Ch-4.md b/mlir/g3doc/Tutorials/Toy/Ch-4.md index 28612fb..1a725db 100644 --- a/mlir/g3doc/Tutorials/Toy/Ch-4.md +++ b/mlir/g3doc/Tutorials/Toy/Ch-4.md @@ -16,7 +16,7 @@ custom optimizations. We show in this chapter how to leverage the Toy Dialect and its high-level semantics to perform transformations that would be difficult in LLVM: first a -simple combine of two redundant operations, and second a full inteprocedural +simple combine of two redundant operations, and second a full interprocedural shape inference with function specialization. # Basic Optimization: Eliminate Redundant Transpose @@ -195,8 +195,8 @@ MLIR also supports `FunctionPass`es which are restricted to modify a single function at a time. This pass couldn't be a function pass due the nature of its interprocedural transformations. -Implementing a such pass is creating a class inheriting from `mlir::ModulePass` -and overriding the `runOnModule()` method: +Implementing such a pass is done by creating a class inheriting from +`mlir::ModulePass` and overriding the `runOnModule()` method: ``` class ShapeInferencePass : public mlir::ModulePass {