sunggg commented on code in PR #14282:
URL: https://github.com/apache/tvm/pull/14282#discussion_r1134582411
##########
tests/python/relax/test_transform_legalize_ops_manipulate.py:
##########
@@ -537,6 +584,54 @@ def reshape(var_rxplaceholder: T.handle, var_T_reshape:
T.handle):
mod = LegalizeOps()(Reshape)
tvm.ir.assert_structural_equal(mod, Expected)
+ # ShapeExpr might be produced by shape computation
+ @tvm.script.ir_module
+ class Reshape2:
+ @R.function
+ def main(x: R.Tensor(("a", "b"), "float32")) -> R.Tensor(("a // 2", "b
* 2"), "float32"):
+ a = T.int64()
+ b = T.int64()
+ lv: R.Shape((a // 2, b * 2)) = R.shape((a // 2, b * 2))
Review Comment:
What do you mean by symbolic dim? If that means rank dynamism, legalization
currently supports.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
