[Mlir-commits] [mlir] [mlir][math] Fix casting mistake in tanh expansion and test (PR #85364)

[llvmlistbot at llvm.org]

srcarroll wrote:

Good point/question. I only knew by observing that sitofp was "wrong" (gave unintentional results) is it gave me -1 for true and 0 for false. Whereas uitofp gives me 1 for true.  But honestly not sure if this works as a target agnostic solution.

https://github.com/llvm/llvm-project/pull/85364