nn: add DenseGeneral generalized linear layer #153381
Open
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Outdated suggestions cannot be applied.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.
Summary
Add
torch.nn.DenseGeneral
, a generalised fully-connected layerthat contracts N input axes (equivalent to
flax.linen.DenseGeneral
)using
torch.tensordot
.Motivation
JAX/Flax and other ecosystems rely heavily on
DenseGeneral
(e.g. parallel attention projections and multi-head outputs).
Adding it to core PyTorch closes a long-standing feature gap,
avoids re-implementations downstream, and lets users stay in
torchscript/compile without custom ops. A reference implementation
has been used in the wild for >1 year :contentReference[oaicite:1]{index=1}.
What’s in this PR
torch/nn/modules/dense_general.py
– new moduletorch/nn/modules/__init__.py
andtorch/nn/__init__.py
API