[Relay] Fix bug in transpose_shape_func (apache#6180)

trevor-m · Aug 26, 2020 · 2273892 · 2273892
1 parent 8b73435
commit 2273892
Show file tree

Hide file tree

Showing 2 changed files with 3 additions and 1 deletion.
diff --git a/python/tvm/relay/op/_transform.py b/python/tvm/relay/op/_transform.py
@@ -540,9 +540,10 @@ def transpose_shape_func(attrs, inputs, _):
     if axes is None:
         axes = list(range(inputs[0].shape[0].value))
         axes.reverse()
+    axes = list(axes)
     for i, axis in enumerate(axes):
         if axis < 0:
-            axes[i] = inputs[0].shape[0] - axis
+            axes[i] = inputs[0].shape[0] + axis
     return [_transpose_shape_func(inputs[0], convert(axes))]
 
 @script

diff --git a/tests/python/relay/test_any.py b/tests/python/relay/test_any.py
@@ -363,6 +363,7 @@ def test_any_transpose():
     verify_any_transpose(any_dims(3), (1, 0, 2), (10, 3, 2))
     verify_any_transpose(any_dims(3), None, (2, 3, 4))
     verify_any_transpose(any_dims(6), (0, 1, 3, 2, 5, 4), (11, 12, 2, 1, 9, 17))
+    verify_any_transpose(any_dims(2), (-1, 0), (3, 2))
 
 def verify_any_squeeze(data_shape, axis, static_data_shape):
     mod = tvm.IRModule()