fix bugs in split converter #402
Merged
Conversation
This file contains hidden or 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
ruoqianguo
commented
Mar 17, 2021
| } | ||
| if (numRemainder) { | ||
| numOutputs += 1; | ||
| sizes.push_back(numRemainder); | ||
| } | ||
| } |
There was a problem hiding this comment.
I only modified the add_split function in select.cpp.
Signed-off-by: Ruoqian Guo <ruoqiang@nvidia.com>
ruoqianguo
force-pushed
the
split_fix_bug
branch
from
f2a9365 to
3990787
Compare
|
Is this really a breaking change? |
narendasan
reviewed
Mar 19, 2021
| @@ -27,10 +27,13 @@ bool add_split(ConversionCtx* ctx, const torch::jit::Node* n, args& args, bool s | |||
| } else { | |||
| auto split_size = args[1].unwrapToInt(); | |||
| numOutputs = inDimSize / split_size; | |||
| if (numOutputs == 1) { | |||
| numRemainder = inDimSize % split_size; | |||
| for (int i = 0; i < numOutputs; i++) { | |||
There was a problem hiding this comment.
style thing: use int64_t if possible
core/conversion/var/Var.cpp
Outdated
| const_layer->setName(("[Freeze Tensor " + tensor_id.str() + " ]").c_str()); | ||
| if (ptr_.ivalue->isTensor()) { | ||
| auto tensor = ptr_.ivalue->toTensor(); | ||
| if (tensor.scalar_type() == at::kLong) { |
There was a problem hiding this comment.
Can we remove this? #407 is adding a formal API to support this.
peri044
reviewed
Mar 19, 2021
Comment on lines
+120
to
+122
| // Split converter generates c10::IValue which hold TensorContainer. | ||
| auto output_container = ptr_.ivalue->toCustomClass<TensorContainer>(); | ||
| out = output_container.get()->tensor(); |
There was a problem hiding this comment.
This looks good to me.
Previously for split, we only considered cases with split as the final output or the list unpack as the final output, which is why the bug occurred for ops that followed them.
…Var.cpp Signed-off-by: Ruoqian Guo <ruoqiang@nvidia.com>
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.
Signed-off-by: Ruoqian Guo ruoqiang@nvidia.com
Description
Fix Split Converter bugs.
core/conversion/converters/select.cpp: add_splitfunction.Var::ITensorOrFreezefunction incore/conversion/var/Var.cpp.In addition, i use the linters to ensure that my code matches the style guidelines. The linters changed the entire file(core/conversion/converters/impl/select.cpp) . I only modified the
add_splitpart in select.cpp.Fixes #401
Type of change
Please delete options that are not relevant and/or add your own.
Checklist: