* Initial changes

* Fix up native functions

* Further fix decoupling

* Remove unnecessary ops...

* Save InputOutputAliases to TorchMlirComputation

* Implement GetResultShape for TorchMlirLow...

* Changes as a result of the LTC TS backend decoupling

* Fix bugs in BackendImpl and codegen
...

* Codegen and build LTC lowering

* Add LazyShapeInference header

* Create ltc_backend.md

* Added introduction and examples

* Added descriptions for files f...

* Added JIT to MLIR lowering

Lowering to JIT is performed in a way similar to how it's done in ...

* Add initial LTC backend skeleton

* Disable CI build and move TorchMLIRPyTorch.cmake

* Update native function definitions

* Add ops to support bert lowering

- Add empty_strided an...

* Xfail unsupported ops

* Register FuncDialect

* Include dynamic_ir in build

* Code reformat
...

* Replace CHECK_EQ with TORCH_CHECK_EQ

* Check value of TORCH_MLIR_USE_INSTALLED_PYTORCH during...

* Remove unnecessary sed in autogen

* Remove .pyc files frrom VCS

* Add default device ordinal API

* Fix reference backend

- Update llvm-project pin to match main

Signed-Off By: Phaneesh Barwaria <[email protected]>

* Update buildAndTest.yml

test with fast-fail matrix builds

* Remove redundant and stateme...

Fixes https://github.com/llvm/torch-mlir/issues/1121

Signed-Off By: Vivek Khandelwal<vivek@nod-...

This commit adds the decomposition for `aten.var.correction` op.

Signed-Off By: Vivek Khandelwa...

This commit adds lowering of `aten.ge.int` op.

Signed-Off By: Vivek Khandelwal <vivek@nod-labs....

This commit fixes the shape calculation for:
1.) aten.mean.dim
2.) aten.var.dim
3.) aten.sum.dim...

- Includes a canonicalizer for `aten.add.t`needed for successfully lowering the shape function
...

* [MHLO] Init MHLO basic Op Conversion
Co-authored-by: Bairen Yi <[email protected]>...

Add source builds and remove deprecated libtorch information.

Follows existing conventions for unary operators.

See RFC https://github.com/llvm/torch-mlir/issues/999

Co-authored-by: Bairen Yi yibairen.byro...

* [MHLO] Add [un]squeeze op patterns

* Conform to llvm coding standard

* minor update

* [MHLO] Init MHLO view like op patterns

See RFC: https://github.com/llvm/torch-mlir/issues/9...

See RFC: https://github.com/llvm/torch-mlir/issues/999

Co-authored-by: Bairen Yi yibairen.byr...

This commit adds verifiers to the ops `ToBuiltinTensorOp` and
`FromBuiltinTensorOp` that make s...

This enables building Pytorch from source in the CI.
The build should mostly hit the ccache.
R...

The unknown dtype case can come from RefineTypes.

This reverts commit ad283c104339267fbe11cc7efefc02a6ad4d2d1e, since it's
causing nightly build ...

In the interest of merging upstream LLVM quickly, a previous patch
(7f08169) updated the torch-...

Co-authored-by: Bairen Yi <[email protected]>
Co-authored-by: Jiawei Wu <xremold@gma...

- Supports cases where the view op expands and collapses dims
simulataneously. This does not ha...

An upstream MLIR bug (that was recently fixed) caused the result to be
ignored for Region- and ...

This patch still only supports a single indexing tensor.

This patch makes some rudimentary changes to torch-mlir's use of MLIR
Python bindings to work w...

This commit adds the support for negative dim cases for `aten.cat`,
`aten.slice.Tensor` and `ate...

This commit adds the canonicalization pattern for the `prim.ListUnpack` op.

Signed-Off By: Vive...

emitError is intended for error cases and not match failures of
patterns. notifyMatchFailure is...

Remove the linking of libtorch/ paths in delocate for CI builds

The biggest change here is to upgrade RefineTypes to the new sparse
dataflow framework.

Smaller...

The original conversion pattern for `AtenBatchNormOp` required that
the input rank be greater t...

This commit adds the decomposition for `aten.var.dim` op.
This commit also make changes in the d...

Remove all the libtorch downloads. If the user sets
-DTORCH_MLIR_USE_INSTALLED_PYTORCH=OFF then...

A previous patch added a new file
("VerifyConversionToValueSemantics.cpp") to the build, but it...

Found while trying to build torch-mlir on an AArch64 Linux VM, worth
a belts and braces to prev...

This patch adds a new pass `torch-verify-conversion-to-value-semantics`,
which looks for non-va...

A previous patch used lowercase names for LIT variables. This patch
replaces them with upperca...

Signed-off-by: Suraj Sudhir <[email protected]>

lowering.

This commit addresses the remaining comments on lowering of
slice_scatter and select_s...

Prior to this patch, the canonicalizers for `AtenSizeOp` and
`AtenSizeIntOp` succeeded only if ...

The way we did it previously still created the slot and copied the
initializer even if unused.

Prior to this patch, the code in the `torch-simplify-shape-calculations`
pass iterated on the u...

This commit does three things:
1. Reverts some of the shape lib changes merged in
https://...

`aten.select_scatter` op.

This commit adds:
1. Lowering of `aten.slice_scatter` op into `tensor...

Set default OSX arch to x86_64. Release builds will override it.
Also update to the latest poin...

A lot of code was super verbose with `torch_mlir.OutputType.XYZ`. Now,
you can simply do `"xyz"`...

The canonicalizer converts `torch.prim.dtype` ops into integer constants
for valid types, but t...

There were some leftovers.

This commit adds support for lists of type `list<optional<tensor>>`
where each element in the l...

Temporarily revert to using PyTorch binaries until source builds
are ready to land.

TORCH_ML...

* Document workaround for PyTorch skew

* Update with review feedback

This patch replaces the use of raw integers like 6, 4, etc. (that
represent PyTorch's scalar ty...

Signed-off-by: Suraj Sudhir <[email protected]>

A prior patch (63538de2) that added support for bfloat16 type did not
add the canonicalization ...

Co-authored-by: Andrew Cain <[email protected]>

On my local machine, `unzip` didn't exist (producing a "command not
found" error), but CMake ig...

Add an option to cache libtorch/ releases if you don't want to
download the latest. Add an opti...

Fixes https://github.com/llvm/torch-mlir/issues/994

* [MLIR][TORCH] Add decomposition of aten.repeat

* refine & rebase

* refine static shapes
...

A previous fix to the handling of size-1 dims in
`aten.view` (https://github.com/llvm/torch-mli...

TorchScript nodes like `prim::Load` and `prim::Store` aren't supported
in torch-mlir because th...

Enables BERT legalization.

Signed-off-by: Suraj Sudhir <[email protected]>

This moves torch-mlir to link against libtorch on macOS and linux

TESTS: Tests pass. Tested r...

- Mention that Bazel build is in peripheral support tier
- Link to the e2e op wiki page

I ran into this while poking around at
https://github.com/llvm/torch-mlir/issues/959

In the `pyhpc_turbulent_kinetic_energy` TorchBench benchmark, the shape
calculation occurs insi...