IntenseWebs/openvino
3
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[IE CLDNN] Switch back to old TI unroller to restore correct matching of TensorIterator output names (#2096)

Browse Source
This commit is contained in:
Mikhail Letavin 2020-09-08 11:48:11 +03:00 committed by GitHub
parent ef3b9e1d1f
commit 671ddeea93
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 2 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
inference-engine/src/cldnn_engine/cldnn_engine.cpp
View File

@ -139,7 +139,8 @@ InferenceEngine::ICNNNetwork::Ptr clDNNEngine::CloneAndTransformNetwork(const In
// Apply all transformations to TensorIterator body
ti_manager.register_pass<ngraph::pass::ApplyTransformationsToTIBody>(manager);
// Unroll will be called after all conversions
ti_manager.register_pass<ngraph::pass::UnrollTensorIterator>();
// temporarily switch back to plugin unroller from NGraph unroller until TI output names are corrected
// ti_manager.register_pass<ngraph::pass::UnrollTensorIterator>();
ti_manager.run_passes(nGraphFunc);
clonedNetwork = InferenceEngine::details::convertFunctionToICNNNetwork(nGraphFunc, *clonedNetwork);