github.com/llvm/torch-mlir commits | Ecosyste.ms: OpenCollective

update PyTorch version to 2.0.0.dev20230113

7f95f4429795d80923a3a9c291ff5a4c77fdd180 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Fix sign-compare warnings in TosaLegalizeCommon.cpp (#1794)

7714bebfe3a5f6d98a4af4831a80a79ad0c671b0 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

update PyTorch version to 2.0.0.dev20230112

f74164161ab524081f514c7f5c1f06d11aa8d955 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

[LINALG][TOSA][MHLO] Add e2e support for aten bitwise ops (#1753)

4f94831fed2aa016f8d3f03b63f8841a064c4152 authored almost 2 years ago by Jiahao Li <[email protected]>

Use global for MLIRTorchOpsIncGenTdFiles (#1784)

6b76e2e927c504898ce9a06c34a5ab0af06ed636 authored almost 2 years ago by Ahmed S. Taei <[email protected]>

update PyTorch version to 2.0.0.dev20230111

cffee681bdf209be7d2cfed75f8c4ee711ef6f1f authored almost 2 years ago by Roll PyTorch Action <[email protected]>

[MLIR][TORCH] Add decomposition for prims.var and prims.sqrt op

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

fd236b2c89158fa8cf4598ab4ca77c82da681f14 authored almost 2 years ago by Vivek Khandelwal <[email protected]>

build: manually update PyTorch version

Set PyTorch and TorchVision version to nightly release 2023-01-08.

Signed-Off By: Vivek Khandelw...

b966733e04868dc0e0b76e300aa09e488f36f228 authored almost 2 years ago by Vivek Khandelwal <[email protected]>

[TOSA] Add LeakyReLU conversion pass (#1790)

* feat(TorchToTOSA): LeakyReLU legalization

* test(LeakyReLU): Add LIT test and enable e2e te...

4e4a571104f355ca41f268ba3f507558aca4183a authored almost 2 years ago by Ashay Rane <[email protected]>

build: update llvm tag to de3f0f7f (#1789)

Credit to @vivekkhandelwal1 for finding the necessary changes.

Summary of changes:

- Swit...

0faba6d2fc01b1e47678ae35054bb79887ffc334 authored almost 2 years ago by Ashay Rane <[email protected]>

Fix unsqueeze in Torch to Tosa conversion (#1780)

0979df6589b532f2d93ecaface98ba912bde143c authored almost 2 years ago by Raghavan Raman <[email protected]>

Added support for aten::norm.ScalarOpt_dim (#1774)

* Added support for aten::norm.ScalarOpt_dim

* Disable NormalizeModule_basic for linalg

c8b867b8760db151727fe9f1b71793d1d1d8a022 authored almost 2 years ago by Gleb Kazantaev <[email protected]>

CI: miscellaneous fixes for Release builds (#1781)

- Use v3 of actions/checkout, since the version we use (v2) uses
Node.js 12, which is deprec...

a897c498036319735f1e718e0c285db75dfae1b3 authored almost 2 years ago by Ashay Rane <[email protected]>

Fix OptionalCType class name (#1779)

* Fix OptionalCType class name

* Rmove LTC xfail tests

8f01072099c7e3001a53716f3a2055d85dfadf6c authored almost 2 years ago by Gleb Kazantaev <[email protected]>

update PyTorch version to 2.0.0.dev20230106

1f2946a5a78c324e4f9e375ddd04dc932d8b6ffa authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20230105

98cb12e3e57f45398064efb3f57728caa6a6f759 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20230104

5351db0e99cd2d3f5f277f52b4135bc5528a0f3c authored almost 2 years ago by Roll PyTorch Action <[email protected]>

build: match windows CI and Release build steps

After a recent LLVM tag update, the Windows Release build failed even
though pre- and post-merge...

6e0b5b9ba740bf19110e58215135b84608c14e0e authored almost 2 years ago by Ashay Rane <[email protected]>

Add e2e support for aten logical or/and/xor/not ops (#1761)

8dc5d985eb637f658725eceb1a81f3f3f4b9af03 authored almost 2 years ago by Jiahao Li <[email protected]>

Add `hasDtype` checks everywhere dtypes are used in decompositions (#1750)

There are several decompositions that assume the operands of the op
have dtypes available; howe...

d44bdd272879200831f170f61369b8828f1aedae authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[custom op] Replace `tanh` dtype function with `expm1` (#1769)

This commit replaces the `tanh` dtype function, which was being used
to test the implementation...

273664ded66a4ad8dcd0bed9540b7684d4cd2074 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

Add E2E support for LeakyRelu and LeakyReluBackward ops (#1733)

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

a88e3766e8c25997dab1b8e8e8b10fb7e7902328 authored almost 2 years ago by Srirammaswamy <[email protected]>

update PyTorch version to 2.0.0.dev20230103

324ccccc2d54f666bd2c94854debfd53ec53ce06 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20230102

818f5dc3add7206f3a92c50260ab1bae83ec848e authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Revert "rebase llvm: 5f24f893cac7aaea292c70f8aa83b021499114be (#1760)" (#1765)

This reverts commit fa356cce50b418269de5a3fa52e95a337cfc78fd.

3d50d3d9fefa62c27fb1989f1efa0b8b1ab987e9 authored almost 2 years ago by powderluv <[email protected]>

update PyTorch version to 2.0.0.dev20230101

cfad718a86e33fddbbe56c80eb0cf58707dd82fa authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20221231

9e71d271f67951cdfb1bca335a665bf10e672472 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

xfail ElementwisePreluModule_basic in dynamo (#1763)

b2dbbba5e6c6b83a2f903c7f72681bf775d911b4 authored almost 2 years ago by powderluv <[email protected]>

rebase llvm: 5f24f893cac7aaea292c70f8aa83b021499114be (#1760)

fa356cce50b418269de5a3fa52e95a337cfc78fd authored almost 2 years ago by Xiafei Qiu <[email protected]>

update PyTorch version to 2.0.0.dev20221230

867fade9c5917baf94d2f69edd015a63897c3098 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Bump to Python 3.8 (#1756)

4dbdc179b7b3eecbdc53685f1e2137c1b485ead0 authored almost 2 years ago by powderluv <[email protected]>

Revert e2e support for aten logical or/and/xor/not ops (#1757)

This reverts commit eaab9be207c6ada9944f63a71b7686f8f5af6bd1, since it
is causing the post-merg...

ac780529b418545a3454c8fb57e27b4ec8b66fbb authored almost 2 years ago by Ashay Rane <[email protected]>

update PyTorch version to 2.0.0.dev20221229

f0998f92dfc0b783a178e2c0250c04d80848130d authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20221228

49274dc3db4b2eee8c7fb72385d23609d01ac21b authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Prelu lowering to linalg (#1712)

Prelu lowering to linalg

2f45959f0d834c3786bde31d93edd4c323418087 authored almost 2 years ago by Shivam Gupta <[email protected]>

update PyTorch version to 2.0.0.dev20221227

ff9523d2b61b3541baac9954a7c0f295ee8d7c22 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20221226

6bd6a79e23c89c08bc5c059b4728c10da2213e3c authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Add e2e support for aten logical or/and/xor/not ops (#1752)

eaab9be207c6ada9944f63a71b7686f8f5af6bd1 authored almost 2 years ago by Jiahao Li <[email protected]>

update PyTorch version to 2.0.0.dev20221225

86d25e310ba21d634a449bad431de039fd9ba215 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20221224

7362f80558dc10a7b6d7c0f7a1635217cf7b67dc authored almost 2 years ago by Roll PyTorch Action <[email protected]>

update PyTorch version to 2.0.0.dev20221223

bcfb1d10b10ca2ba0a894eca0f08ea14189f7268 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Allow passing traced `torch.nn.Module`s into `torch_mlir.compile` (#1743)

This commit adds support for passing to `torch_mlir.compile` the
result of running `torch.jit.t...

3260a1ea6e885a2b19dbcb391ea22c5cbf742047 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

update PyTorch version to 2.0.0.dev20221222

52669cbbd5973d537a7d23720822a3b45dd07495 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

[MHLO] Evaluate RuntimeAssertOp at compile time (#1732)

49071f86e650363a1edf27cd5613845296013886 authored almost 2 years ago by Jiahao Li <[email protected]>

[Bazel] Fix Bazel build

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

ddbcf569e0ecbb0b0608ad18dbef1209c949adca authored almost 2 years ago by Vivek Khandelwal <[email protected]>

Revert "reimplement linear lowering torchToMhlo (#1524)" (#1744)

This reverts commit 50b524546f389b7f6243780f4332e077388dd512.

297fd3aa47e7bd13dc307f04ef455f39c8c4d885 authored almost 2 years ago by Tanyo Kwok <[email protected]>

Add aten.std.correction op and its decomposition (#1731)

60a139271d6e71a8d314cad45575e28041b23a2c authored almost 2 years ago by Jiahao Li <[email protected]>

reimplement linear lowering torchToMhlo (#1524)

50b524546f389b7f6243780f4332e077388dd512 authored almost 2 years ago by zzp_miracle <[email protected]>

[Torch][MHLO] Decompose aten.copy op. Lower aten.rsqrt & sigmoid to mhlo. (#1734)

15b249777bceb0d7f036af2eb993aa91622b6f75 authored almost 2 years ago by Jiahao Li <[email protected]>

[TOSA] Add aten.gather support for tosa (#1680)

9dc09ac8c54dabadf703c37f7d3cf21486412d9e authored almost 2 years ago by Chi_Liu <[email protected]>

[TOSA] Add aten.masked_fill.Tensor/Scalar support (#1735)

b2cefc0b642a316946b6368a7f1e444e32b90dae authored almost 2 years ago by Chi_Liu <[email protected]>

update PyTorch version to 2.0.0.dev20221221

810473cc03a5d3dcd34121349c7ed86c55129990 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

Added GeluBackward: MHLO support (#1725)

0f6008c80252cf11a49e595842596f3f2740c27f authored almost 2 years ago by pranavmulticore <[email protected]>

Unrevert #1724 (#1737)

* Unrevert #1724

* Update pytorch requirements.txt

1d695239ff8d4fc9e185f2a45d4a25a8f878a0bb authored almost 2 years ago by Jae Hoon (Antonio) Kim <[email protected]>

[RefineTypes] Fix knowledge dtype for `aten.embedding` op

-- The dtype of the result of `aten.embedding` should match that of
the `weight` operand's (o...

66d7a412cbbbd85c2256c09c12efb3168887114a authored almost 2 years ago by Abhishek Varma <[email protected]>

build: update llvm tag to 7ccbb4df (#1736)

Summary of changes:

- LLVM now includes <optional> instead of "llvm/ADT/Optional.h" in most
...

577e38da58fd1f25997480685a83e6ba975f9f31 authored almost 2 years ago by Tanyo Kwok <[email protected]>

ci: post comment on RollPyTorch tracker issue upon build failure (#1730)

Now that the RollPyTorch tracker issue exists, we can automate the job
of notifying folks of fa...

f6b6069a343ee640a91e9cabb884a01e052bc0c5 authored almost 2 years ago by Ashay Rane <[email protected]>

update PyTorch version to 2.0.0.dev20221217

335cfe913446d53e75d96d99a9c986e9c268b17a authored almost 2 years ago by Roll PyTorch Action <[email protected]>

build: fix LTC code after upstream PyTorch change (#1727)

pytorch/pytorch@140a3139 reverted a change from yesterday, causing the
RollPyTorch action to br...

dd1cf578a6e2aab2e0b7442d97c27ee1bd034514 authored almost 2 years ago by Ashay Rane <[email protected]>

[TORCH] Add Complex Number support (#1673)

Add Complex number dtype support to torch tensors. Add
aten.fft_fft op to test complex numbers.

17ee643aeb03f08517d2e86c11f91764a2fe5ec1 authored almost 2 years ago by ataheridezfouli-groq <[email protected]>

Replace asIntArrayRefSlow with macro (#1724)

* Replace asIntArrayRefSlow with macro

* Update pytorch requirements.txt

a2a93891ea365012ebd2bc461ee45bfac1764bba authored almost 2 years ago by Jae Hoon (Antonio) Kim <[email protected]>

Add `report_fatal_error` to `getTypeForScalarType` (#1722)

Functions like `getTypeForScalarType` that do a mapping from one set
of types to another should...

211cf8fc361f44a943d2827a33e1369f1ac0dc36 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

Pass op legality info to `verifyBackendContractPass` (#1705)

In order to verify if a given IR satisfies the backend contract, the
verifier needs to know if ...

60db793feb63c5e95112d6bdf64cf431605fe4ca authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

Yapf Format `refbacked.py`.

8ba77ae2a519bc8db9f5c0ae9197471845b925cf authored almost 2 years ago by Prashant Kumar <[email protected]>

Add float16 support in the refbackend.

This will require https://reviews.llvm.org/D139121 patch to go through.

564403e3a1ba0415ffe00117bf4289ffad7e44d5 authored almost 2 years ago by Prashant Kumar <[email protected]>

Update buildAndTest.yml (#1723)

cd90c0aaf5ef3f9a07f32d1a9650e2efdcc1bae1 authored almost 2 years ago by powderluv <[email protected]>

[torchdynamo] Move to aot_autograd instead of raw make_fx

As [@ezyang suggested](https://github.com/pytorch/pytorch/issues/90276#issuecomment-1339791275),...

af9e8a5e63e2ca8d498d781d9d56f2c2f8362146 authored almost 2 years ago by Sean Silva <[email protected]>

ci: print ccache statistics and configuration at end of CI run (#1719)

There appear to be two problems with the caching layer in our CI runs:
(a) the sizes of some of...

64f9a0e978d530a2c00d2134e6a01c41cb034b9a authored almost 2 years ago by Ashay Rane <[email protected]>

update PyTorch version to 2.0.0.dev20221214

a29f173a6b94c09a597a3397d26d450b51d02c0d authored almost 2 years ago by Roll PyTorch Action <[email protected]>

[cleanup] Fix a few more llvm::None -> std::nullopt

b60da34f8498b43859676c32c48d690c533130bf authored almost 2 years ago by Sean Silva <[email protected]>

Minor fixes for development.md

- Mention the rotation doc
- Fix minor typos / broken link

8c3774bb2a270bb73a8e38040f70dc13feb3b1d6 authored almost 2 years ago by Sean Silva <[email protected]>

build: update llvm tag to 3a020527 (#1717)

Summary of changes:

- Replace `llvm::None` with `std::nullopt`, since the former is deprecat...

f63bb9f86c27adef85392d845c42e4cd212005be authored almost 2 years ago by Ashay Rane <[email protected]>

Add aten.slice.Tensor & aten.cat folders (#1691)

b1f68328497c53aa2a9aba5dfa079c69a0b72132 authored almost 2 years ago by Ahmed S. Taei <[email protected]>

ci: run `git pull` before committing pytorch version updates (#1716)

The RollPyTorch action often takes more than 1.5 hours to finish.
During this time, if another ...

731c3132311020a984b3ea8f2cc3a1c9041924e2 authored almost 2 years ago by Ashay Rane <[email protected]>

Disable pypi publishing.

See https://github.com/llvm/torch-mlir/issues/1709

07a65961dd43b895addc1694b4a44b5be3d7457e authored almost 2 years ago by Daniel Ellis <[email protected]>

[custom op] Generalize shape library logic to work with dtypes (#1594)

* [custom op] Generalize shape library logic to work with dtypes

This commit generalizes the ...

a710237437337a59cf067accab2c24a90fbb1d48 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[README] Small touch-ups, and mention PT2

2acf7da63c9d7deaeacbe79096619620dee4539e authored almost 2 years ago by Sean Silva <[email protected]>

update PyTorch version to 2.0.0.dev20221213

8d098dc8d5e28186fff26efb6781f6fbeb707415 authored almost 2 years ago by Roll PyTorch Action <[email protected]>

[TOSA] Add aten.add/sub.Scalar/Tensor si64 type support (#1604)

163d19cce6ba1d97269f948438da206e38f53f6e authored almost 2 years ago by Chi_Liu <[email protected]>

Make `getTensorRank` safer by changing return to `Optional<unsigned>` (#1707)

Currently `getTensorRank` returns -1 if it was unable to get the rank
of the tensor. However, n...

73bd32d06cffc92dd4056a1a0e41595d85ccb451 authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[cleanup] fix naming of private variable according to the style guide (#1704)

430737b8202a27b97347101196e1c061b9bec7bb authored almost 2 years ago by Ashay Rane <[email protected]>

[cleanup] Use `"` instead of `'` for string literals

This is the more predominant style in the codebase. I'm sure there are
more in other parts of th...

a5959420330ceb6a83ca373cd352f3e28c62681f authored almost 2 years ago by Sean Silva <[email protected]>

[MLIR][TORCH] Add e2e support for aten.var_mean op

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

d4862ec611ae728e3ed41ccbcab86e9946bd21c7 authored almost 2 years ago by Vivek Khandelwal <[email protected]>

build: manually update PyTorch version

Set PyTorch and TorchVision version to nightly release 2022-12-11.

Signed-Off By: Vivek Khandel...

143a8f378d8d35c0ea81c521c07db2a03be411a1 authored almost 2 years ago by Vivek Khandelwal <[email protected]>

Revert "[MLIR][TORCH] Fix mean and mean.dim op for large-sized inputs"

This reverts commit 55c7e66aa7e3df407652d796280d02144b8ea899.

f783e19dcb377bb3d48d721484a537901aca5c7c authored almost 2 years ago by Vivek Khandelwal <[email protected]>

Remove eager_mode

This was an experimental attempt at rolling out own op-by-op executor
with `__torch_dispatch__`,...

7731211d02200127e97e259163df911f046d1fbd authored almost 2 years ago by Sean Silva <[email protected]>

[CI] Verify bazel buildifier is run and changes committed (#1700)

Ensures the buildifier (linter for bazel build files) is run and changes are pushed.

109c91ae9b7f47c64dfe94d8e95f7dda08243fb7 authored almost 2 years ago by Sambhav Jain <[email protected]>

Extended TorchMLIRLoweringContext with virtual CreateComputation method (#1699)

* Extended TorchMLIRLoweringContext with virtual CreateComputation method

* Fix device_data_c...

804f9f1f8f2002c3f1537e1fb0c919e30f6b4600 authored almost 2 years ago by Gleb Kazantaev <[email protected]>

[Bazel] Resolve circular dependency and add targets for conversion to MLProgram dialect (#1694)

A circular dependency was introduced in https://github.com/llvm/torch-mlir/commit/e7edcc62fde300...

f8a2592905bdbe5b3d5cba42f17fda4ec76648f7 authored almost 2 years ago by Sambhav Jain <[email protected]>

Allow running DecomposeComplexOps more than once (#1671)

The current implementation of `DecomposeComplexOps` fails if an op
expected to be decomposed do...

a54b334578d4bbbaad3408fd671ba01ff6e03f2f authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[cleanup] Use a single function pipeline for TOSA->Linalg

This should run faster and is overall clearer.

e8511840c3330f3fe2925ad843650fc273b610be authored almost 2 years ago by Sean Silva <[email protected]>

Remove unnecessary decompose-complex-ops tests (#1693)

This commit removes lit tests from the `decompose-complex-ops` that
are essentially testing a m...

76190e8a3fc868b3ce2d22849b7c464af7eb592f authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[RefBackend] Add elementwise fusion and buffer deallocation

This gives some decent improvements to memory consumption and latency of
testing. I would have e...

69171c246ab7855149657842ece27e10aaa8e4ac authored almost 2 years ago by Sean Silva <[email protected]>

[e2e tests] Rename default config from "refbackend" to "linalg"

This more accurately reflects what it is. The previous name was
conflating the use of RefBackend...

29c88234642857cfca50ade05915bd719923ce36 authored almost 2 years ago by Sean Silva <[email protected]>

build: update llvm tag to 798fa4b4 (#1684)

- Support for non-prefixed accessors has been removed. See:
https://reviews.llvm.org/D136727
...

dd35488da53f099b8920af705dea7c234d89627f authored almost 2 years ago by Ramiro Leal-Cavazos <[email protected]>

[torchdynamo] Add ResNet18 example with TorchDynamo

This is a minor variation on our other resnet18 examples swapping in
TorchDynamo.

We replicate ...

b1f9e09f8537021cc9d87628bd8de110796c071a authored almost 2 years ago by Sean Silva <[email protected]>

Publish releases to PyPI after build

98d80a642aa3bab53ba134f2517bfe81d87bb36f authored almost 2 years ago by Daniel Ellis <[email protected]>

[cleanup] Remove disabled e2e test

This test has been disabled a long time, and since RefBackend is so slow
we don't want to add th...

c956c39c8603ff4859f95c1a5299c9f42ba0d77d authored almost 2 years ago by Sean Silva <[email protected]>

[docs] Add info about special e2e testing cases.

d52359a89147103657a1f014dcbb83c9984d7243 authored almost 2 years ago by Sean Silva <[email protected]>

[MLIR][TORCH] Add E2E support for randn and randn.generator op

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

3e4bb2bd8ed18a61ae1d31542f516e6f5b104771 authored almost 2 years ago by Vivek Khandelwal <[email protected]>