Skip to content

Add hlo_convolution() op#161

Open
sebffischer wants to merge 1 commit into
mainfrom
dev/20260428-convolution
Open

Add hlo_convolution() op#161
sebffischer wants to merge 1 commit into
mainfrom
dev/20260428-convolution

Conversation

@sebffischer
Copy link
Copy Markdown
Collaborator

Implements the StableHLO convolution operation per spec (https://openxla.org/stablehlo/spec#convolution): generic N-D windowed convolution with strides, padding, lhs/rhs dilations, window reversal, feature/batch grouping, and a precision_config. Layout is described via a ConvDimensionNumbers struct that renders to MLIR's #stablehlo.conv<...x...->...> syntax.

Spec constraints C1–C27 are checked in infer_types_convolution(), each annotated with its constraint id. Quantized-tensor constraints (C28–C34) are skipped because the package does not yet model quantized types.

Implements the StableHLO convolution operation per spec
(https://openxla.org/stablehlo/spec#convolution): generic N-D windowed
convolution with strides, padding, lhs/rhs dilations, window reversal,
feature/batch grouping, and a precision_config. Layout is described via
a `ConvDimensionNumbers` struct that renders to MLIR's
`#stablehlo.conv<...x...->...>` syntax.

Spec constraints C1–C27 are checked in `infer_types_convolution()`,
each annotated with its constraint id. Quantized-tensor constraints
(C28–C34) are skipped because the package does not yet model
quantized types.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant