Layer Normalization using the torch.aten.native_layer_norm

https://github.com/nod-ai/SHARK-T...

maps onnx.expand to torch aten broadcast_to, three tests added

---------

Co-authored-by: K...

Set PyTorch and TorchVision version to nightly release 2024-01-09.

Signed-Off By: Vivek Khandelw...

Signed-Off By: Vivek Khandelwal <[email protected]>

Signed-Off By: Vivek Khandelwal <[email protected]>

Signed-Off By: Vivek Khandelwal <[email protected]>

Adaptive pooling ops can only be decomposed into their non-adaptive
counterparts in trivial cas...

Floating-point literals should use the correct type specifier.

The lowering for `aten.clamp` did not support integer types. Added
support for integer types in...

Co-authored-by: hanhanW <[email protected]>

The doc seems copy-and-paste from the linalg-on-tensors class

fixes #2711

Set PyTorch and TorchVision version to nightly release 2024-01-02.

Signed-Off By: Vivek Khandel...

Need this for LayerNorm

Adds a lowering to Linalg for reflection_pad1d. Based on ideas/code from draft PR
https://githu...

Co-authored-by: Rob Suderman <[email protected]>

(instead of an input)

Addresses part of #2689. fixes #2702

https://github.com/llvm/torch-mlir/issues/2646

Decompose aten.exponential() into: -exp(1-x)/l...

As noted in the plan when this work started, we need to produce an ORT
EP plugin for a downstre...

lower onnx min op to torch aten minimum op

Addresses two of the ops from
https://github.com/llvm/torch-mlir/issues/2689

https://github....

This commit adds the OnnxToTorch support for GlobalAveragePool op.

Signed-Off By: vivekkhande...

This commit adds the OnnxToTorch support for Reshape op.

Signed-Off By: Vivek Khandelwal <[email protected]>

This commit adds the OnnxToTorch support for BatchNormalization and Concat op.

Signed-Off By: v...

Some names for `onnx` identifiers are not legal in `mlir-ir`. Sanitize
so that the generated `i...

Changes made during upstreaming:

* Removed comments attributing some copied code back to torc...

The expression for HardSigmoid in Onnx
(https://onnx.ai/onnx/operators/onnx__HardSigmoid.html):...

- Going through the `#torch-mlir` channel on the `llvm` discord, I
realize that there are some ...

This commit adds the OnnxToTorch support for Conv and ConvTranspose op.

Signed-Off By: vivekkha...

This PR adds the `enable_ir_printing` option to `torch_mlir.compile`,
which can be used to prin...

Started work on the `tan` lowerings for ONNX to Torch. Uses `sin` and
`cos` to represent a `tan`.

Broadcasting for `torch.aten.matmul` is optional so a MxN with NxK
matmul should be legalized t...

Adding the `--progress` flag shows the same output as what `git clone`
would show. This is very...

https://github.com/llvm/torch-mlir/commit/be3e74b647375d1c72f075bebf77dc31e74477a2
breaks bazel...

TorchOnnxToTorch
For Erf function

Modify PYTHONPATH to new related directory in docs.

This commit adds the OnnxToTorch support for AveragePool op.

Signed-Off By: vivekkhandelwal14...

This commit adds the OnnxToTorch support for ReduceSum, ReduceMean, and
ReduceMin ops.

This commit adds the OnnxToTorch support for Gelu op.

---------

Co-authored-by: Rob Suderm...

Adapts bazel build per https://github.com/llvm/torch-mlir/pull/2648.

https://github.com/sj...

General lowering for `onnx.Gemm` to `torch`

The three remaining compare operations
onnx.Greater
onnx.Less
onnx.GreaterOrEqual

Are al...

Adds a lowering from `onnx.sinh` to `aten.sinh`. This includes adding
the `aten.sinh` operator.

Lowerings for `transpose` from ONNX to `aten`. Implementation depends on
making multiple `aten....

This replaces the lowering of aten.cat with tensor.concat, allowing more
efficient handling of ...

Includes the lowering from the `aten` equivalent to `tensor` operations.

This commit adds support for `LeakyRelu and GatherElements` op in the
onnx pipeline.

Signed-...

This bumps the llvm submodule to HEAD to pick up recent fixes.

This commit adds the OnnxToTorch support for Reciprocal, Round,
ScatterElements, Sigmoid, Sin, ...

Added the less or equal operation to OnnxToTorch.
onnx.LessOrEqual

---------

Co-authored...

Lowerings for `selu` lowerings for ONNX to the corresponding torch
implementations. Torch's `se...

Add `aten.selu` operation to `torch` dialect.

This commit adds the OnnxToTorch support for Matmul.

Simple Python console script to import an ONNX protobuf to the torch
dialect for additional pro...

Co-authored-by: LiuYuanqiang <[email protected]>

This is part 1 of 2, which will also include upstreaming the FX
importer. I started with ONNX b...

For easier tracking of issues, sort the TOSA passing list. It is still
significantly smaller th...

This depends on a change in the LLVM core repository which adds acos
support to the MLIR Math di...

`AtenStackOp` defines this folder for list operand containing single
element:
```
OpFoldResul...

This commit adds the OnnxToTorch support for BitwiseXor, BitwiseOr, Div, Equal, Cast,
Ceil, Floo...

As title, support torch.aten.einsum op

Right now only support Static Shape, because of the kn...

Signed-Off By: Vivek Khandelwal <[email protected]>

The linalg Op `linalg.conv_2d_ngchw_fgchw` had a bug where

1. Weights were accessed as G,F,C,...

llvm-project: bbd2b08b95fe76bea138c1b03c1cd42ed3ee04df
stablehlo: ab709fe48de88c67717abfbd7ef17...

MSVC-19 does not support constexprs of lambda captured constexpr values
like this: https://godb...

Despite aten.mm requiring the input and output types match, we still opt
to maintain signedness...

Adds a lowering for the torch.aten.argmin operator to linalg via decomposition into torch.aten.m...

Try to fix the error reported by @qingyunqu in #2609.

Fix #2608

Set PyTorch and TorchVision version to nightly release 2023-12-04.

Signed-Off By: Vivek Khandelw...

Some of docs referred to old file paths that no longer exists. This
patch updates some of the i...

The function `getTypeForScalarType` currently takes an argument to
specify the signedness of in...

Bump LLVM to `5e5a22caf88ac1ccfa8dc5720295fdeba0ad9372` and StableHLO to
`83f095e7217c897f1ecca...

decompose like:
```
import torch

def my_triu(x, diag):
rows = torch.ops.aten.size(x, -...

Adapts to the TorchOnnxToTorch changes from
https://github.com/llvm/torch-mlir/pull/2585.
Also...

This commit adds the OnnxToTorch support for Atan, Bitshift, BitwiseAnd,
and BitwiseNot op.
This...

Prior to this, the concurrency rules for presubmits (which cancel eagerly) were being applied to...

This was unfortunately being initialized in a directory below its first use. This was causing th...

The aten.reshape ops in the decomposition are replaced with prims.collapse
and prims.split_dim...

Adds a pipeline to convert custom ops and metadata represented as
`torch.operator` custom ops t...