Skip to content

Add constant weights for GEMMs to refitmap #556

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
kevinch-nv merged 1 commit into from
Nov 4, 2020
Merged

Add constant weights for GEMMs to refitmap #556

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions builtin_op_importers.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1062,6 +1062,8 @@ DEFINE_BUILTIN_OP_IMPORTER(Gemm)
{
auto transposedWeights = ctx->createTempWeights(weights.type, weights.shape);
ASSERT(transposeWeights(weights, {1, 0}, &transposedWeights), ErrorCode::kUNSUPPORTED_NODE);
transposedWeights.setName(weights.getName());
LOG_WARNING("Weight " << transposedWeights.getName() << " has been transposed! If you plan on overwriting this weight with the Refitter API, the new weights must be pre-transposed");
weights = transposedWeights;
// Since we've already transposed now, we can set transpose to false.
transB = false;
Expand All @@ -1070,6 +1072,7 @@ DEFINE_BUILTIN_OP_IMPORTER(Gemm)
= ctx->network()->addConstant(weights.shape, static_cast<nvinfer1::Weights>(weights));
// Map the constant layer to the weights name.
ctx->registerLayer(weightsLayer, node.input(1));
ctx->insertRefitMap(weights.getName(), weightsLayer->getName(), nvinfer1::WeightsRole::kCONSTANT);
inputB = weightsLayer->getOutput(0);
}
else
Expand Down
⚡