masahi commented on a change in pull request #10691:
URL: https://github.com/apache/tvm/pull/10691#discussion_r831643114
##########
File path: tests/python/frontend/tensorflow/test_forward.py
##########
@@ -2421,10 +2422,11 @@ def _test_sparse_to_dense(sparse_indices,
sparse_values, default_value, output_s
)
oshape = tf.constant(output_shape, shape=output_shape.shape,
dtype=str(output_shape.dtype))
+ # Output shape depends on a dynamic input, use VM.
if default_value == None:
output = tf.sparse_to_dense(indices, oshape, values)
compare_tf_with_tvm(
- [sparse_indices, sparse_values], ["indices:0", "values:0"],
output.name
+ [sparse_indices, sparse_values], ["indices:0", "values:0"],
output.name, mode="vm"
Review comment:
This test was previously working on the graph runtime just because the
dynamic input is bound to a constant tensor before `relay.build(...)` in TF
tests. Such usage of dynamic inputs makes no sense so I just changed it to use
vm for testing.
##########
File path: tests/python/frontend/tensorflow/test_forward.py
##########
@@ -2421,10 +2422,11 @@ def _test_sparse_to_dense(sparse_indices,
sparse_values, default_value, output_s
)
oshape = tf.constant(output_shape, shape=output_shape.shape,
dtype=str(output_shape.dtype))
+ # Output shape depends on a dynamic input, use VM.
if default_value == None:
output = tf.sparse_to_dense(indices, oshape, values)
compare_tf_with_tvm(
- [sparse_indices, sparse_values], ["indices:0", "values:0"],
output.name
+ [sparse_indices, sparse_values], ["indices:0", "values:0"],
output.name, mode="vm"
Review comment:
This test was previously working on the graph runtime just because the
dynamic input is bound to a constant tensor before `relay.build(...)` in TF
tests. Such usage of dynamic inputs makes no sense so I just changed it to use
vm for testing.
--
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]
