diff options
author | Alexandre Passos <apassos@google.com> | 2018-10-08 13:50:12 -0700 |
---|---|---|
committer | TensorFlower Gardener <gardener@tensorflow.org> | 2018-10-08 13:58:40 -0700 |
commit | eec9ca8f0baccd249a49046fe31b460903e44850 (patch) | |
tree | b6397af544af7c05abca4bea08bd6354f90bedf1 /tensorflow/core/common_runtime | |
parent | 494bbdfced3fd8596721d12e73676c4967f452e4 (diff) |
Partial support tfe.defun in tf.gradients.
Doesn't attempt to deal with cases where we might have already generated
the functiondef for the parent function as in that case we cannot easily
modify the forward pass.
PiperOrigin-RevId: 216243224
Diffstat (limited to 'tensorflow/core/common_runtime')
-rw-r--r-- | tensorflow/core/common_runtime/shape_refiner.cc | 5 |
1 files changed, 5 insertions, 0 deletions
diff --git a/tensorflow/core/common_runtime/shape_refiner.cc b/tensorflow/core/common_runtime/shape_refiner.cc index fa4d1eda62..9488a44778 100644 --- a/tensorflow/core/common_runtime/shape_refiner.cc +++ b/tensorflow/core/common_runtime/shape_refiner.cc @@ -288,6 +288,11 @@ Status ShapeRefiner::SetShape(const Node* node, int output_port, "output_port '", output_port, "' is out of range, ", "node '", node->name(), "' has ", node->num_outputs(), " outputs"); } + // Note: it's possible, if the node's been updated, that the shape inference + // context doesn't have the right number of outputs. + if (node->num_outputs() > c->num_outputs()) { + TF_RETURN_IF_ERROR(c->ExpandOutputs(node->num_outputs())); + } // Check compatibility, and merge the shapes. ShapeHandle existing_shape = c->output(output_port); |