Handle optional inputs and remove more empty shape nodes in TensorRT EP #3455
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jywu-msft
reviewed
Apr 8, 2020
onnxruntime/core/providers/tensorrt/tensorrt_execution_provider.cc
Outdated
Show resolved
Hide resolved
skottmckay
reviewed
Apr 9, 2020
| const auto name = input.second->Name(); | ||
| if (!name.empty()){ | ||
| meta_def->inputs.push_back(name); | ||
| } | ||
| } |
skottmckay
reviewed
Apr 9, 2020
| meta_def->inputs.push_back(input.second->Name()); | ||
| const auto name = input.second->Name(); | ||
| if (!name.empty()){ | ||
| meta_def->inputs.push_back(name); |
There was a problem hiding this comment.
checking input.second->Exists() may be clearer/simpler.
skottmckay
reviewed
Apr 9, 2020
| for (const auto& tensor : init_tensors) { | ||
| graph_build.AddInitializedTensor(*(tensor.second)); | ||
| } | ||
|
|
||
| ORT_ENFORCE(graph_build.Resolve().IsOK()); |
jywu-msft
approved these changes
Apr 10, 2020
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
When TRT EP partitions a graph for TRT, optional input of operators become subgraph's input, which will cause problem since it doesn't have name/type/values. This PR removed optional inputs from subgraph input list.
TRT doesn't support empty shape in current release. When a model has Nonzero op, there is a chance that Nonzero's output has empty shape during runtime. This mostly happened in RCNN models. As a workaround this PR excluded empty shape related nodes from TRT subgraph. New TRT release 7.1 will fix the issue and support empty shape.