llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
torch-mlir/projects/pt1/e2e_testing/xfail_sets.py

4867 lines
194 KiB
Python
Raw Normal View History

Dual license the torch-mlir project. This commit (with approval from all contributors) dual licenses the torch-mlir project under both the standard LLVM license and the standard PyTorch license. This will facilitate moving code between torch-mlir and the two upstream projects. The standard file comment is now: ``` // This file is licensed under the Apache License v2.0 with LLVM Exceptions. // See https://llvm.org/LICENSE.txt for license information. // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception // Also available under a BSD-style license. See LICENSE. ``` See `LICENSE` in the project root for the terms of both licenses.
2021-09-30 00:03:40 +08:00
# Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
# See https://llvm.org/LICENSE.txt for license information.
# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
# Also available under a BSD-style license. See LICENSE.
Add IREE support in TorchScript e2e tests. - Add support for "expected failures" in test reporting. The new error reports look like [this](https://gist.github.com/silvasean/6ffd95e1d55302b699673da201da210d). - We will now be able to put these tests into CI, since the harness understand which tests are expected to pass and fail. - Refactor RefBackendTestConfig to NpcompBackendTestConfig which supports both RefBackend and IREE. - Add instructions for installing IREE dependencies (both from packages and for local builds of IREE) - Add `tools/torchscript_e2e_test.sh` for invoking the e2e test harness (this makes invoking a bit easier, as it doesn't rely on a loose Python invocation).
2021-07-01 05:13:21 +08:00
# This file describes the sets of tests expected to fail for each config.
# This information is deliberately kept in a side table, rather than
# in-situ on the test, as a deliberate layering decision: tests should
# have unique keys to identify them and enable side tables of various kinds
# (this includes down into lower parts of the stack, where a side table
# might be used to keep more elaborate sets of testing configurations).
Move COMMON_TORCH_MLIR_LOWERING_XFAILS into test_suite That way, downstreams don't have to duplicate this list. Also, remove "external config" feature, since it is subsumed by just importing the test suite.
2022-04-20 03:35:56 +08:00
from torch_mlir_e2e_test.test_suite import COMMON_TORCH_MLIR_LOWERING_XFAILS
Add make_fx_tosa variant to end2end tests (#2240) * Add make_fx_tosa variant to end2end tests * e2e_testing/xfail_sets.py: Add make_fx_tosa xfail for stable
2023-07-13 21:07:54 +08:00
from torch_mlir._version import torch_version_for_comparison, version
Move COMMON_TORCH_MLIR_LOWERING_XFAILS into test_suite That way, downstreams don't have to duplicate this list. Also, remove "external config" feature, since it is subsumed by just importing the test suite.
2022-04-20 03:35:56 +08:00
Advance llvm-project and stablehlo. (#2619) llvm-project: bbd2b08b95fe76bea138c1b03c1cd42ed3ee04df stablehlo: ab709fe48de88c67717abfbd7ef17425eb95ddaf These commits were chosen in order to account for an MLIR API break from https://github.com/llvm/llvm-project/commit/3dbac2c007c114a720300d2a4d79abe9ca1351e7 which required a patch to stablehlo. We integrate a bit beyond that commit to deal with some revert/reapply cycles in the intervening range which were discovered in another downstream. Further, it requires adaptation to the stablehlo API breaks introduced from https://github.com/openxla/stablehlo/pull/1872 which are along for the ride. Since some stablehlo builders were changed to directly take int64_t array refs, also traced that up some call stacks to eliminate some signed/unsigned mismatches that result. Also adds a few TOSA tests to the passing set that seem to work now.
2023-12-08 15:13:42 +08:00
print(f"TORCH_VERSION_FOR_COMPARISON =", torch_version_for_comparison())
[TOSA] Fix conversion for depthwise convolutions (#2398) * [TOSA] Fix conversion for depthwise convolutions * Add e2e tests for depthwise and grouped convolutions Co-authored-by: Lucas Camphausen <lucas.camphausen@iml.fraunhofer.de>
2023-08-18 23:15:54 +08:00
LINALG_XFAIL_SET = COMMON_TORCH_MLIR_LOWERING_XFAILS | {
Modifies onnx resize lowering to fix numerical issues (#3381) Updates: - some unsupported modes are now going to report a match failure for unsupported coordinate transformation modes. - fixes a bug that was introduced in the last patch for resize (my bad...) - uses actual x and y coordinates for computing weights in bilinear interpolation (rather than eps modified values) - slightly simplifies the bilinear interpolation payload for readability and performance - passes coordinate transformation mode information from an onnx.Resize op to the mode string for the aten._interpolate op. This allows us to perform custom logic in the torch->linalg lowering to support onnx.Resize options without losing the default behaviors of the interpolate op.
2024-05-31 08:34:37 +08:00
# lowering to torch backend IR fails due to unsupported op: aten.upsample_[mode/dims].vec
# these interpolate tests are added specifically to test onnx.Resize.
"InterpolateDynamicModule_sizes_bilinear",
"InterpolateDynamicModule_sizes_nearest",
"InterpolateStaticModule_scales_bilinear_align_corners",
"InterpolateDynamicModule_scales_recompute_bilinear",
[Linalg] Promote type for compare tensor op (#3416)
2024-06-05 07:05:39 +08:00
"ElementwiseFloatTensorGtIntTensorModule_basic",
[stablehlo] add aten::_int_mm lowering (#3474) as title
2024-06-20 16:10:31 +08:00
"AtenIntMM_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
# unimplemented lowering torch -> linalg for torchvision.deform_conv2d
# this is added to check the torch.onnx.export -> import_onnx -> torch path
"DeformConv2D_basic",
[Stablehlo] support aten.any.dim, aten.min.dim (#3500) * refactor `TorchToStablehlo/Reduction.cpp` * add `ConvertAtenReduceWithIndicesOp` patterns
2024-06-29 16:53:33 +08:00
"ReduceAnyDimFloatModule_basic",
[TOSA] Fix Tensor.hacked_twin to support diff size indexes (#3547) - Broadcasts index list tensors - Adds torch.nn.Unfold test Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-07-31 05:32:05 +08:00
"UnfoldModule_basic",
[TOSA] Fix conversion for depthwise convolutions (#2398) * [TOSA] Fix conversion for depthwise convolutions * Add e2e tests for depthwise and grouped convolutions Co-authored-by: Lucas Camphausen <lucas.camphausen@iml.fraunhofer.de>
2023-08-18 23:15:54 +08:00
}
E2E framework: Add support for list/dict/scalar values Most of the change is in the reporting code to give error messages that are useful, and adjusting TraceItem to be semantically correct w.r.t. Python's modeling of return values. This allows writing a test like `ListLiteralModule_basic` for list functionality, which we will soon be hooking up to IREE. The IR for that test currently gets this far: ``` builtin.func @forward(%arg0: f64) -> !torch.list<!torch.float> { %0 = torch.from_f64 %arg0 %1 = torch.prim.ListConstruct %0, %0 : (!torch.float, !torch.float) -> !torch.list<!torch.float> return %1 : !torch.list<!torch.float> } ``` It should be sufficient to just add a conversion of `torch.prim.ListConstruct` (+ relevant type conversion) to necessary IREE primitives. For lists of *tensors* (rather than scalar floats), it gets more complicated, as we need to deal with changing their element type to ValueTensorType first (by default, they will all be NonValueTensorType). It seems that IREE might have a type we can lower into for non-value tensors as well, TBD.
2021-08-11 07:10:31 +08:00
build: manually update PyTorch version (#3627) Set PyTorch and TorchVision version to nightly release 2024-08-18. This commit also updates the `scaled_dot_product_attention` op. A new attribute `enable_gqa` has been added. As of now, only the default value for the same is supported. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-19 14:33:56 +08:00
if torch_version_for_comparison() < version.parse("2.5.0.dev"):
LINALG_XFAIL_SET = LINALG_XFAIL_SET | {
# Error: 'torch.aten.scaled_dot_product_attention' op expected 8 operands, but found 7
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
# WORKS FOR TORCH VERSION 2.5.0.dev20240902, REMOVE WHEN ENABLE_GQA IS PUT IN STABLE
"ScaledDotProductAttentionBoolMaskModule_basic",
"ScaledDotProductAttentionDifferentCausalModule_basic",
build: manually update PyTorch version (#3627) Set PyTorch and TorchVision version to nightly release 2024-08-18. This commit also updates the `scaled_dot_product_attention` op. A new attribute `enable_gqa` has been added. As of now, only the default value for the same is supported. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-19 14:33:56 +08:00
"ScaledDotProductAttentionDifferentModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
"ScaledDotProductAttentionMaskModule_basic",
"ScaledDotProductAttentionSameCausalModule_basic",
"ScaledDotProductAttentionSameDynamicModule_basic",
build: manually update PyTorch version (#3627) Set PyTorch and TorchVision version to nightly release 2024-08-18. This commit also updates the `scaled_dot_product_attention` op. A new attribute `enable_gqa` has been added. As of now, only the default value for the same is supported. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-19 14:33:56 +08:00
"ScaledDotProductAttentionSameModule_basic",
}
[torch] Add support for `torch.view` with dynamic shapes (#3164) We can map to `tensor.reshape` for handling multiple output dynamic shapes. Later we can perform a more complex analysis for indentifying expand/collapse cases from the tensor.reshape. Initially we planned to handle this identification at the `torch` level however it will be easier to handle once converted to core mlir-dialects.
2024-04-19 02:47:19 +08:00
LINALG_CRASHING_SET = {
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
# Runtime op verification: Out of bounds access
"AtenDiagEmbedNegOffsetDiag_basic",
"AtenDiagEmbedNonDefault4DDiag_basic",
"AtenDiagEmbedOffsetDiag_basic",
"AtenDiagEmbedRevDimDiag_basic",
"AtenEmbeddingBagStaticModule_basic",
"AtenEmbeddingBagSumExample_basic",
"Aten_EmbeddingBagExample_basic",
# Runtime op verification: subview is out-of-bounds of the base memref
"Conv_Transpose1dModule_basic",
"Conv_Transpose1dStaticModule_basic",
"Conv_Transpose2dModule_basic",
"Conv_Transpose2dStaticModule_basic",
"Conv_Transpose3dModule_basic",
"Conv_Transpose3dStaticModule_basic",
"ConvolutionModule2DTransposeStridedStatic_basic",
"ConvolutionModule2DTransposeStrided_basic",
"GridSamplerBasic1_basic",
"GridSamplerBasic2_basic",
"GridSamplerBasic3_basic",
"GridSamplerBasic4_basic",
# Runtime op verification: stride mismatch in memref.cast
"ReduceAllDimEmpty_basic",
"TraceUnsignedIntModule_empty",
"TraceModule_empty",
[torch] Add support for `torch.view` with dynamic shapes (#3164) We can map to `tensor.reshape` for handling multiple output dynamic shapes. Later we can perform a more complex analysis for indentifying expand/collapse cases from the tensor.reshape. Initially we planned to handle this identification at the `torch` level however it will be easier to handle once converted to core mlir-dialects.
2024-04-19 02:47:19 +08:00
# Crashes due to copy to a smaller destination buffer than the source buffer.
"SliceCopyStartGreaterThanDimSize_Module_basic",
}
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
TORCHDYNAMO_XFAIL_SET = {
#### General TorchDynamo/PyTorch errors
[torch] Add folders for `torch.fill`, `torch.ones`, `torch.zeros` and `aten.getItem` (#2849) So that the CumSum Op in OPT can get the constant that it requires to be lowered to TMTensor --------- Co-authored-by: Rob Suderman <rob.suderman@gmail.com> Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
2024-02-03 02:46:33 +08:00
# torch._dynamo.exc.Unsupported: Tensor.item
"CumsumModule_basic",
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
# TypeError: new_empty(): argument 'size' (position 1) must be tuple of ints, but found element of type NoneType at pos 0
# RuntimeError: Failed running call_function aten.convolution_backward(...
[torchdynamo] Update XFAIL sets with upstream bug numbers.
2022-11-24 22:36:13 +08:00
# https://github.com/pytorch/pytorch/issues/89629
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2D_basic",
[torch-mlir][sparse] add sparsification to linalg reference backend (#2887) This adds a few passes that will ensure linalg with sparse tensors are properly lowered to loops and can run using the ExecutionEngine for testing (a few details on parameter passing from PyTorch still TBD) Test results: $ ./tools/e2e_test.sh --config linalg Summary: Passed: 1144 Expectedly Failed: 8 $ python -m e2e_testing.main --config=torchdynamo -v Summary: Passed: 960 Expectedly Failed: 163 Filed issue: https://github.com/pytorch/pytorch/issues/119407
2024-02-09 01:37:31 +08:00
# Size result mismatch (exposed by downstream canonicalizer
# on incompatabile casts).
# https://github.com/pytorch/pytorch/issues/119407
"ConvolutionBackwardModule2DStrided_basic",
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
# RuntimeError: Index tensor must have the same number of dimensions as self tensor
# RuntimeError: Failed running call_function aten.nll_loss_backward(...
[torchdynamo] Update XFAIL sets with upstream bug numbers.
2022-11-24 22:36:13 +08:00
# https://github.com/pytorch/pytorch/issues/89630
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
[torchdynamo] Use decompositions to support a few ops
2022-11-29 22:01:42 +08:00
# TypeError: uniform() missing 2 required keyword-only arguments: 'dtype' and 'device'
# RuntimeError: Failed running call_function aten.uniform(...
# https://github.com/pytorch/torchdynamo/issues/1954
"UniformNoCorrelationModule_basic",
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
#### Torch-MLIR internal compiler errors
# These are probably due to slightly different ops being recorded by
# torchdynamo vs. torchscript.
[torchdynamo] Use decompositions to support a few ops
2022-11-29 22:01:42 +08:00
# No upstream decompositions.
[torchdynamo] Annotate the XFAIL's with more info
2022-11-25 20:33:34 +08:00
# %6:4 = torch.operator "aten._embedding_bag_forward_only"(%1, %3, %5, %false, %int0, %false, %none, %false, %int-1) : (!torch.tensor<*,f32>, !torch.tensor<*,si64>, !torch.tensor<*,si64>, !torch.bool, !torch.int, !torch.bool, !torch.none, !torch.bool, !torch.int) -> (!torch.tensor, !torch.tensor, !torch.tensor, !torch.tensor)
[torchdynamo] Use decompositions to support a few ops
2022-11-29 22:01:42 +08:00
# See also: https://github.com/pytorch/torchdynamo/issues/327
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
"AtenEmbeddingBagSumExample_basic",
xfail ElementwisePreluModule_basic in dynamo (#1763)
2023-01-01 05:54:25 +08:00
# error: unsupported by backend contract: tensor with unknown rank
# note: see current operation: %1 = "torch.tensor_static_info_cast"(%arg0) : (!torch.vtensor<[5,4,3,2,1],f32>) -> !torch.vtensor<*,f32>
"ElementwisePreluModule_basic",
[Torch] Support prelu decomposition (#3069)
2024-03-29 08:05:00 +08:00
# error: torch._dynamo.exc.BackendCompilerFailed: backend='compiler_fn' raised: AssertionError: Unregistered operation: torch.aten._prelu_kernel
"ElementwisePreluStaticModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
# ERROR: value (Tensor with shape=[2, 3, 6, 10], dtype=torch.float32, min=-1.336e-32, max=+0.9152, mean=+0.4837) is not close to golden value (Tensor with shape=[2, 3, 6, 10], dtype=torch.float32, min=+0.02233, max=+0.9152, mean=+0.4777)
build: manually update PyTorch version Set PyTorch and TorchVision version to nightly release 2023-02-06. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-02-07 13:38:44 +08:00
"UpSampleNearest2dDynamicFactor_basic",
[MLIR][TORCH] Add support for integer type input for max.dim op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-02-06 19:52:04 +08:00
"ReduceMaxAlongDimUnsignedInt_basic",
Add aten.min.dim to linalg lowering (#2600)
2023-12-05 23:16:35 +08:00
"ReduceMinAlongDimUnsignedInt_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
# ERROR: value (-56) is not equal to golden value (200)
build: manually update PyTorch version Set PyTorch and TorchVision version to nightly release 2023-02-20. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-02-20 15:40:09 +08:00
"AtenIntTensorByteDtypeModule_basic",
Fix handling of non-int tensors in `getScalarValue` (#1914) The current implementation of `getScalarValue` does not check that the input to a `ValueTensorLiteralOp` is an i64 before extracting the value, and it does not check that the result type of the `PrimNumToTensorScalarOp` is also an i64. This leads to crashes or invalid IR generated when the `input` is something other than an i64 tensor or `!torch.int`. This commit addresses those issues. In addition, the function `getScalarValue` is renamed to `getScalarIntValue` to make it clear that it *only* extracts scalar integers.
2023-03-07 02:12:58 +08:00
# ERROR: assert isinstance(e, FakeTensor)
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
[Torch Dialect] add RSub, ScalarImplicit canonicalize (#1899) * add rsub, scalarimplit canonicalizer * reformat * address comments * fix bug * fix test * Update elementwise.py * resolve merge conflict * change to 3 * change to 3 * real fix * fix name * add torchdynamo fail test --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-03-07 09:38:27 +08:00
# ERROR: assert isinstance(e, FakeTensor)
"RsubInt0d_NumToTensor_Module_basic",
[MLIR][TORCH] Add decomposition for prims::squeeze op This commit adds the decomposition for the prims.squeeze op. Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-11-16 13:57:58 +08:00
# ERROR: RuntimeError: Found a custom (non-ATen) operator that either mutates or its inputs: prims::squeeze.. Getting these operators to work with functionalization requires some extra work. For mutable ops you need to register a corresponding out-of-place variant of the op, and you also need to register a Functionalization kernel that performs some boilerplate, telling functionalization to map from the mutable op to the out-of-place op. See a more complete example of how to do this at https://gist.github.com/bdhirsh/7dadbf6296f8f7d1abcf4c482f438aaa.
"PrimsSqueezeModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
Lowering to linalg of prims split_dim op (#2576) Adds support for lowering to prims split_op. Similar design to collapse op lowering in https://github.com/llvm/torch-mlir/pull/2572, with some small differences, because the split_dim op (in pytorch) is view-changing whereas the collapse is not. The difference means that 1) it must be registered in the function Torch::isViewLikeOp 2) it must be be added to the "expected fail" set for the torch dynamo backend.
2023-11-21 23:56:09 +08:00
"SplitDimStaticModule_basic",
"SplitDimDynamicModule_basic",
[MLIR][TORCH] Add support for prims::view_of op This op does nothing and just returns the input operand as the result of the op. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-04-10 11:50:26 +08:00
# ERROR: RuntimeError: Found a custom (non-ATen) operator that either mutates or its inputs: prims::view_of.. Getting these operators to work with functionalization requires some extra work. For mutable ops you need to register a corresponding out-of-place variant of the op, and you also need to register a Functionalization kernel that performs some boilerplate, telling functionalization to map from the mutable op to the out-of-place op. See a more complete example of how to do this at https://gist.github.com/bdhirsh/7dadbf6296f8f7d1abcf4c482f438aaa.
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
[Dynamo] turn on `no_python=True` for dynamo tests (#2040)
2023-04-29 07:05:17 +08:00
# See https://github.com/llvm/torch-mlir/pull/2040 and corresponding upstream issue
# https://github.com/pytorch/pytorch/issues/99752.
# torch._dynamo.exc.Unsupported: call_function BuiltinVariable(bool) [TensorVariable()] {}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# START tests failing due to: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(float) [TensorVariable()] {}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenSubFloatModule_basic",
"AtenMulFloatModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"CeilFloatModule_basic",
"DivFloatModule_basic",
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GtFloatIntModule_basic",
"NeFloatIntModule_basic",
"SubFloatModule_basic",
"MulFloatModule_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(float) [TensorVariable()] {}
# START tests failing due to: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(int) [TensorVariable()] {}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenIntTensorCharDtypeModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"DivIntModule_basic",
"EqIntModule_basic",
"GeIntModule_basic",
"GtIntModule_basic",
"MulIntModule_basic",
"NeIntModule_basic",
"SqrtIntModule_basic",
"SubIntModule_basic",
"TensorToIntZeroRank_basic",
"TensorToInt_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(int) [TensorVariable()] {}
[torch] Support lowering `torch.item` to `tensor.extract` (#2835) Extracting scalar values from tensors can be implemented via a lowering to tensor.extract.
2024-02-01 07:09:12 +08:00
# ERROR: torch._dynamo.exc.Unsupported: Tensor.item
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenItemIntOpModule_basic",
"AtenItemFpOpModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: call_method ListVariable() sort [] {'reverse': ConstantVariable(bool)}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"SortIntListReverse_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: call_method ListVariable() sort [] {}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"SortIntList_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# START tests failing due to: torch._dynamo.exc.Unsupported: data dependent operator: aten._local_scalar_dense.default
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenFloatScalarModule_basic",
"AtenIntBoolOpModule_basic",
"QuantizedMLP_basic",
"QuantizedSingleLayer_basic",
"QuantizedBatchedInputSingleLayer_basic",
"QuantizedNoLayer_basic",
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: torch._dynamo.exc.Unsupported: data dependent operator: aten._local_scalar_dense.default
# START tests failing due to: torch._dynamo.exc.Unsupported: dynamic shape operator: aten.bincount.default
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: torch._dynamo.exc.Unsupported: dynamic shape operator: aten.bincount.default
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor bool call_function aten.Bool
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"BoolFloatConstantModule_basic",
"BoolIntConstantModule_basic",
add uncovered cases in view lowering (#2524) removes unecessary checks from empty strided
2023-11-02 10:56:44 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.size
"ViewSizeFromOtherTensor_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor bool call_function aten.__contains__
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ContainsIntList_False",
"ContainsIntList_True",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor bool call_function aten.all
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor bool call_function aten.any
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor float call_function aten.sqrt
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"SqrtIntConstantModule_basic",
[Torch-to-Linalg] Add dynamic dimension support for BroadcastTo op (#2174) -- This commit adds support for dynamic dimension in BroadcastTo op. Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
2023-07-08 01:01:51 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.size
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"BroadcastDynamicDimModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# START tests failing due to: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.Int
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"IntFloatModule_basic",
"PowIntFloatModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.Int
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.len
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"LenStrModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function aten.numel
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"NumelModule_basic",
"NumelZeroRankModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function prim.max
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"PrimMaxIntModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: torch._dynamo.exc.Unsupported: torch.* op returned non-Tensor int call_function prim.min
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"PrimMinIntModule_basic",
"PrimMinIntDynamicModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# START tests failing due to: empty graph in dynamo
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"IsFloatingPointFloat_True",
"IsFloatingPointInt_False",
"TorchPrimLoopForLikeModule_basic",
"TorchPrimLoopWhileLikeModule_basic",
[Importer] import constant tuple (#2132) * [Importer] import constant tuple * update * update * update
2023-05-31 14:14:14 +08:00
"ScalarConstantTupleModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# END tests failing due to: empty graph in dynamo
# ERROR due to: backend never runs because of empty frame
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ConstantBoolParameterModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# START tests failing due to: 'torch.aten.mul.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
"UpSampleNearest2dDynamicSize_basic",
"UpSampleNearest2dStaticFactor_basic",
"UpSampleNearest2dStaticSize_basic",
"UpSampleNearest2d_basic",
# END tests failing due to: 'torch.aten.mul.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
# START tests failing due to: 'torch.aten.add.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
"ElementwiseAddScalarFloatModule_basic",
# END tests failing due to: 'torch.aten.add.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
# ERROR: 'torch.aten.add.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.int'
"ElementwiseAddScalar_TensorLiteralInt32_Module_basic",
"HBC_basic",
# ERROR: 'torch.aten.div.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
"ElementwiseDivScalarModule_basic",
fix(TorchToTosa.cpp): adjust torch->tosa div conversion (#2200) check the return type of the division to figure out whether to use the floating point implementation of a division or to use the integer. the issue rose from the fact that the inputs are all integer but the result was casted to floating point. The conversion then chose to use the integer implementation of division which is not legal in tosa when all the inputs get casted to floating point. fix(TorchToLinalg): AtenDivScalarOp upcast self operand as well if applicable, the self operand must also be casted to float as it can be an integer.
2023-06-12 17:18:38 +08:00
# ERROR: 'torch.aten.div.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.int'
"ElementwiseAtenDivIntScalarModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: 'torch.aten.sub.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.float'
"ElementwiseSubScalarFloatModule_basic",
"ElementwiseSubScalarIntModule_basic",
# ERROR: Exception: Unsupported: missing default value for argument 0 in schema for aten.div.Tensor_mode
Added 2 Ops: Floor divide scalar and Floor divide scalar mode (#3156) - Added linalg lowering for `AtenFloorDivideScalarOp` - Needed `AtenDivScalarModeOp` for the decomp. - Added linalg lowering for `AtenDivScalarModeOp` - Moved linalg payload logic to `createDivModePayload()` since the logic was nearly identical for both `AtenDivScalarModeOp` and `AtenDivTensorModeOp`. Just a template function - Added `AtenDivScalarModeOp` lowering for stablehlo Pytorch's [`torch.floor_divide()`](https://pytorch.org/docs/stable/generated/torch.floor_divide.html) in a previous version (for a reason unknown to me) preformed a truncation instead of "floor". The already implemented op `AtenFloorDivideTensorOp` was done before this change. However, this wasn't caught because our testcases only tested positive floor division. I changed this to floor as well as adding a few test cases.
2024-04-16 04:45:10 +08:00
"ElementwiseAtenFloorDivideScalarNegativeModule_basic",
"ElementwiseAtenFloorDivideScalarModule_basic",
"ElementwiseDivScalarRoundingModeFloorModule_basic",
"ElementwiseDivScalarRoundingModeTruncModule_basic",
"ElementwiseDivScalarRoundingModeFloorStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncStaticModule_basic",
"ElementwiseDivScalarRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncIntStaticModule_basic",
[dynamo] Verify the default value is passed by kwargs (#2998)
2024-04-28 02:18:33 +08:00
# ERROR: 'torch.aten.mul.Tensor' op operand #1 must be Any Torch tensor type, but got '!torch.int'
[RFC] general support for Adaptive Pooling Ops (#2661) Adaptive pooling ops can only be decomposed into their non-adaptive counterparts in trivial cases. For example, the current decomposition for AtenAdaptiveAvgPool1dOp in DecomposeComplexOps.cpp supports outSize = inSize (i.e., do literally nothing), and outSize = 1 (i.e., do a batched average). The reason adaptive pooling ops are difficult to lower to linalg is that they are not constantly strided. They are computed by taking an input tensor of shape (N, C, Hin), and an output size Hout, and computing the output tensor at position (n,c, h) in the following way: 1. compute st(h) = (h*Hin)//Hout 2. compute en(h) = 1 + ((h+1)*Hin -1)//Hout 3. apply a computation (max or avg) to the slice: INPUT[n, c, st(h):en(h)] The provided sample implementation (for ConvertAtenAdaptiveAvgPool1dOp) uses tensor.extract to access the input tensor inside the payload of a linalg generic op. This is likely an unattractive use of linalg generic ops, which is why I am asking for some more targeted feedback on the validity of this approach before attempting to support the many other adaptive pooling ops. Specifically: - Is the performance of this implementation bad enough to warrant targeting different dialects entirely? e.g. TMtensor/linalg ext/ etc. - If the provided implementation is of acceptable performance to the community, then is it permissable to remove the Adaptive pooling decompositions from DecomposeComplexOps.cpp? Based on the current structure of the -torch-decompose-complex-ops pass, it does not seem possible to only decompose the adaptive ops in special cases (it seems to get stuck in an infinite loop on a match failure). I would be happy to instead incorporate the case logic into the conversion directly, and remove the decompositions once they are rendered completely obsolete. As long as this approach is acceptable, I can clean up the implementation with some helper functions, and quickly add support for each of the remaining Adaptive pooling ops.
2024-01-10 03:14:10 +08:00
"AdaptiveAvgPool1dStaticLargerOutput_basic",
"AdaptiveAvgPool1dGeneralDynamic_basic",
Converts all Adaptive Pooling Ops to Linalg (#2808) The previous conversions for AtenAdaptiveAvgPool1dOp and AtenAdaptiveMaxPool2dOp are refactored into a general templated conversion that works for all of the AtenAdaptive...PoolNdOp's. New support is added for the following ops: 1. AtenAdaptiveMaxPool1d 2. AtenAdaptiveMaxPool3d 3. AtenAdaptiveAvgPool3d Support is also provided for passing inputs without batch dimensions. For example, applying adaptive_avg_pool2d to an input tensor of rank 3. After [pytorch #118162](https://github.com/pytorch/pytorch/pull/118162) gets down to torch-mlir, I'll add a test for AdaptiveMaxPool1d with return_indices (which will pass with that upstream fix). --------- Co-authored-by: James Newling <james.newling@gmail.com>
2024-03-23 02:05:20 +08:00
"AdaptiveAvgPool1dGeneralDynamicNoBatches_basic",
"AdaptiveAvgPool2dDynamic_basic",
"AdaptiveAvgPool2dDynamicNoBatch_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# ERROR: Exception: Unsupported op: get_attr
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"TensorFloatModule_basic",
"TensorIntModule_basic",
# START tests failing due to: complex floating point ops
# END tests failing due to: complex floating point ops
[Torch Op] Add unbind.int support with ListUnpack (#2058) * add unbind int * reformat * use unpack canonicalize * address comments * Empty commit, trigger test * add ltc blacklist * clean up * address comments * check permute list * erase in recompose --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-19 10:07:58 +08:00
# ERROR: Exception: Unsupported: return type List[Tensor] in schema for aten.unbind.int
"UnbindIntListUnpack_Module_basic",
"UnbindIntGetItem_Module_basic",
[TM_TENSOR] Add `aten.scatter.[src|value]` op This commit adds support of `aten.scatter.src` and `aten.scatter.value` ops. Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-10-16 05:46:06 +08:00
# ERROR: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(float) [TensorVariable()] {}
"ScatterValueFloatModule_basic",
# ERROR: torch._dynamo.exc.Unsupported: call_function BuiltinVariable(int) [TensorVariable()] {}
"ScatterValueIntModule_basic",
[MLIR][TORCH] Add support for aten._unsafe_index_put.hacked_twin op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-07-14 15:26:54 +08:00
# AssertionError: Unregistered operation: torch.aten._unsafe_index_put
"UnsafeIndexPutHackedTwin1DFloatNonAccumulateModule_basic",
[MLIR][TORCH] Add StableHLO lowering for embedding_bag.padding_idx op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-09-05 21:28:37 +08:00
# AssertionError: Unregistered operation: torch.aten._embedding_bag_forward_only
"AtenEmbeddingBagStaticModule_basic",
[MLIR][TORCH] Add support for conversion to int8 dtype Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-09-29 20:19:18 +08:00
# Lowering not present for this case
"ElementwiseToDtypeI64ToUI8Module_basic",
[MLIR][TORCH] Add support for int8 dtype for sub, add, and bitwise_and op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-10-03 19:59:56 +08:00
# torch._dynamo.exc.TorchRuntimeError: Failed running call_function <built-in method add of type object at 0x7f4f8b05a720>(*(FakeTensor(..., size=(3, 4), dtype=torch.int8), 3, 2), **{}): Tensor with dtype torch.int64 is not the expected dtype of torch.int8!
"ElementwiseAddScalarInt8Module_basic",
build: manually update PyTorch version (#2521)
2023-10-19 22:03:00 +08:00
# ERROR: dtype (torch.int64) is not equal to golden dtype (torch.float32)
"ThresholdBackward2dMixedModule_basic",
[Torch Dialect] support AtenArangeStartOutOp in ReduceOpVariants like… (#2563) … AtenBernoulli_FloatOp It fixing case like: `%2110 = torch.aten.arange.start_out %int1, %int1517, %int1, %2109 : !torch.int, !torch.int, !torch.int, !torch.tensor -> !torch.tensor`. `aten.arange.start_out` doesn't have value semantics also, means`%2110` is an alias for %2109. So I decompose it to `aten.arange.start` + `torch.contents.overwrite`. The complex decomposition logic is target to handle cases like view and dtype cast which I add in e2e tests.
2023-11-17 00:51:55 +08:00
# ERROR: shape (torch.Size([12])) is not equal to golden shape (torch.Size([3, 4]))
"ArangeStartOutViewModule_basic",
[torch][quant] Support quantize and dequantize for torch (#2731) Handle both `torch.dequantize` and `torch.quantize_per_tensor` including the op based quantization parameter tracking. This includes adding `qint32` to torch types as it was missing during the initial type inclusion. For testing we only have `torch.int8` and `torch.float` types on function boundaries as the `qint8` types require passing the scale and zero point quantization information which is not supported yet.
2024-01-13 11:11:14 +08:00
# Dynamo does not support tracing quantized tensors
[torch] `torch.dequantize` for per channel tensors to` linalg` (#2769) Support a lowering for dequantization for per channel tensors from `torch` dialect to a linalg decomposition. Tested via a numerical `torch` test.
2024-01-26 08:40:21 +08:00
"ElementwiseDequantizePerChannelModule_basic",
[torch][quant] Support quantize and dequantize for torch (#2731) Handle both `torch.dequantize` and `torch.quantize_per_tensor` including the op based quantization parameter tracking. This includes adding `qint32` to torch types as it was missing during the initial type inclusion. For testing we only have `torch.int8` and `torch.float` types on function boundaries as the `qint8` types require passing the scale and zero point quantization information which is not supported yet.
2024-01-13 11:11:14 +08:00
"ElementwiseDequantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorModule_basic",
[torch] Fix clamp ranges on quantize_per_tensor on unsigned (#3018) SExtValue was used for `int` and `uint` clamp values. This caused the result to always be outputed as `zero`.
2024-03-21 04:37:47 +08:00
"ElementwiseQuantizePerTensorUIntModule_basic",
[torch][quant] Quantized `torch.mm` for linalg with end-to-end test (#2750) This includes custom op matching for decomposed operations and fusing dequantization into dense operations. As a validation we compare to the dequant+mm torch implementation.
2024-01-25 06:02:50 +08:00
"AtenMmQuint8_basic",
Adds Some Quantization Support for AtenMatmulOp (#3147) 1. onnx.MatMulInteger now converts to aten.matmul instead of aten.mm 2. aten.matmul, for ranks >=2, now allows quantized inputs and will lower to linalg::quantized_matmul or linalg::quantized_batch_matmul. 3. added AtenMatmulOp to the FuseQuantizeOps rewrite patters QuantizeOperands, QuantizeTransposedOperands, and QuantizeAccumulator 4. added several tests, including some to test AtenMmOp with varying quantization signed-ness. 5. a quantized matmul mat-vec test is added to verify the failure to lower to linalg; cleaned of out-of-date code related to common torch-mlir lowering xfails. 6. in debugging a real model with quantized matmuls, I found a bug on the scalarize-shapes pass which resulted from the aten.full op folder returning an incompatible result type. This is fixed by the small change here to [lib/Dialect/Torch/IR/TorchOps.cpp](https://github.com/llvm/torch-mlir/compare/main...zjgarvey:torch-mlir:MatMulIntegerFix?expand=1#diff-dc8ed165c207918e606490eee3984b1ad51d7034e6aac36fc046bf47f6f03f4f).
2024-04-16 07:06:47 +08:00
"AtenMmQint8_basic",
"AtenMmQMixedSigni8_basic",
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
[TorchToLinalg] Adds Support for Remaining Quantized Matmul Cases (#3167) The new cases added for quantized matmuls are: 1. vec-vec 2. vec-mat 3. mat-vec each of which are now lowered to expand(s), quantized_matmul, and collapse.
2024-04-17 00:28:28 +08:00
"AtenMatmulQint8VV_basic",
"AtenMatmulQint8VM_basic",
Adds Some Quantization Support for AtenMatmulOp (#3147) 1. onnx.MatMulInteger now converts to aten.matmul instead of aten.mm 2. aten.matmul, for ranks >=2, now allows quantized inputs and will lower to linalg::quantized_matmul or linalg::quantized_batch_matmul. 3. added AtenMatmulOp to the FuseQuantizeOps rewrite patters QuantizeOperands, QuantizeTransposedOperands, and QuantizeAccumulator 4. added several tests, including some to test AtenMmOp with varying quantization signed-ness. 5. a quantized matmul mat-vec test is added to verify the failure to lower to linalg; cleaned of out-of-date code related to common torch-mlir lowering xfails. 6. in debugging a real model with quantized matmuls, I found a bug on the scalarize-shapes pass which resulted from the aten.full op folder returning an incompatible result type. This is fixed by the small change here to [lib/Dialect/Torch/IR/TorchOps.cpp](https://github.com/llvm/torch-mlir/compare/main...zjgarvey:torch-mlir:MatMulIntegerFix?expand=1#diff-dc8ed165c207918e606490eee3984b1ad51d7034e6aac36fc046bf47f6f03f4f).
2024-04-16 07:06:47 +08:00
"AtenMatmulQint8_basic",
[torch] Adds Quantization Support for `aten.relu` (#3177) A choice was made to quantize the return type of Relu with a scale and zero point copied from the input's quantization scheme. With this choice, the torch-to-linalg conversion of quantized Relu essentially computes max(input, zeroPoint) in the elementwise payload.
2024-04-24 02:01:36 +08:00
"QuantizedReluInt32_basic",
"QuantizedReluInt8_basic",
"QuantizedReluUint8_basic",
[torch] Support `torch.convolution` quantized lowering to `linalg` (#2811) Linalg has quantized specific operations. We can lower to these operations when there is a known zeropoint and scale operations. This allows the `convolution` to occur with lower bitwidth's, improving the overall performance.
2024-01-31 05:46:47 +08:00
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[TorchToLinalg] Adds Quantization Support for ConvTranspose (#3240) I spent a little while debugging numerics issues with some tests similar to the ones in quantized_models.py, only to find that pytorch's quantized conv transpose is catastrophically inaccurate. I'll upstream the issue and only leave the tests here which are of the form quantize -> dequantize -> op.
2024-05-01 00:23:09 +08:00
"ConvTranspose2DQInt8_basic",
implement aten.conv1d, aten.conv3d, and aten.conv_tbc (#2757) convolution with [time,batch,channel] ordering, as opposed to the default [batch, channel, time]. Currently implementing by transposing the input and output, but may need to get its own implementation in the future because this is supposed to be an op that gives a speedup. This is used by fairseq (https://github.com/facebookresearch/fairseq/issues/172). (in case you were wondering like me, this is different from transposed convolution. Transposed convolution has fractional strides). --------- Co-authored-by: Xida Ren <xida.ren.dev@gmail.com> Co-authored-by: Frederik Harwath <frederik.harwath@amd.com>
2024-01-24 13:30:03 +08:00
# Dynamo not supporting conv_tbc
"ConvTbcModule_basic",
[MLIR][TORCH] Extend support for OnnxToLinalg lowering for Dropout and Div op (#2938) Fixes https://github.com/nod-ai/SHARK-Turbine/issues/451, https://github.com/nod-ai/SHARK-Turbine/issues/452
2024-02-27 13:32:05 +08:00
"FloatImplicitModule_basic",
"IntImplicitModule_basic",
[onnx][torch] Gridsampler E2E test and corrections of gridsampler (#2987) The addition of an e2e test is actually provided in the Shark-Testsuite. This adds 2 test cases for the gridsampler e2e test. Also as intended there were some items found which needed correction, so the Gridsampler op has also a change.
2024-03-07 02:56:58 +08:00
# Others
[dynamo] Verify the default value is passed by kwargs (#2998)
2024-04-28 02:18:33 +08:00
"ExponentialModule_basic",
[onnx][torch] Gridsampler E2E test and corrections of gridsampler (#2987) The addition of an e2e test is actually provided in the Shark-Testsuite. This adds 2 test cases for the gridsampler e2e test. Also as intended there were some items found which needed correction, so the Gridsampler op has also a change.
2024-03-07 02:56:58 +08:00
"GridSamplerBasic1_basic",
"GridSamplerBasic2_basic",
[dynamo] Verify the default value is passed by kwargs (#2998)
2024-04-28 02:18:33 +08:00
"GridSamplerBasic3_basic",
[Torch Dialect] support aten.fake_quantize_per_tensor_affine (#3014)
2024-03-15 08:53:29 +08:00
"FakeQuantizePerTensorAffineModule_basic",
"FakeQuantizePerTensorAffineDynamicShapeModule_basic",
"FakeQuantizePerTensorAffineRoundToEvenModule_basic",
[Dynamo] turn on `no_python=True` for dynamo tests (#2040)
2023-04-29 07:05:17 +08:00
}
TORCHDYNAMO_CRASHING_SET = {
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
# No upstream decompositions.
# %6:4 = torch.operator "aten._embedding_bag_forward_only"(%1, %3, %5, %false, %int0, %false, %none, %false, %int-1) : (!torch.tensor<*,f32>, !torch.tensor<*,si64>, !torch.tensor<*,si64>, !torch.bool, !torch.int, !torch.bool, !torch.none, !torch.bool, !torch.int) -> (!torch.tensor, !torch.tensor, !torch.tensor, !torch.tensor)
# See also: https://github.com/pytorch/torchdynamo/issues/327
"Aten_EmbeddingBagExample_basic",
"FullModuleInt3D_basic",
"ThresholdBackward1dIntModule_basic",
"ThresholdBackward2dIntModule_basic",
"ThresholdBackward3dIntModule_basic",
# See https://github.com/llvm/torch-mlir/issues/2050
[Dynamo] turn on `no_python=True` for dynamo tests (#2040)
2023-04-29 07:05:17 +08:00
"ElementwiseCloneChannelsLastMemoryFormatModule_basic",
[Dynamo] fix TORCHDYNAMO_CRASHING_SET (#2082)
2023-05-02 10:35:26 +08:00
"ElementwiseCloneContiguousModule_basic",
"ElementwiseCloneModule_basic",
[Dynamo] turn on `no_python=True` for dynamo tests (#2040)
2023-04-29 07:05:17 +08:00
"ExpandAsFloatModule_basic",
"ExpandAsIntModule_basic",
"ExpandModule_basic",
"MoveDimIntModule_basic",
"MoveDimIntNegativeIndexModule_basic",
[Dynamo] fix TORCHDYNAMO_CRASHING_SET (#2082)
2023-05-02 10:35:26 +08:00
"NarrowVerticalTest2_basic",
"NarrowVerticalTest_basic",
[Dynamo] turn on `no_python=True` for dynamo tests (#2040)
2023-04-29 07:05:17 +08:00
"NumpyTRank2Module_basic",
"NumpyTRankNDynamicModule_basic",
"NumpyTRankNStaticModule_basic",
"PermuteModule_basic",
"PermuteNegativeIndexModule_basic",
"SelectIntNegativeDimAndIndexStaticModule_basic",
Pure-Python FX importer. (#2098) Co-authored-by: Sean Silva <silvasean@google.com>
2023-05-12 13:46:33 +08:00
"TestMultipleTensorAndPrimitiveTypesReturn_basic",
[Dynamo] fix TORCHDYNAMO_CRASHING_SET (#2082)
2023-05-02 10:35:26 +08:00
"TModuleRank2_basic",
"ToCopyModule_basic",
"TransposeIntModule_basic",
"TransposeIntNegDimsModule_basic",
[TOSA] Add aten._index_put_impl support (#2031) Add e2e support by add "tosa-to-scf"
2023-07-18 00:51:24 +08:00
"IndexPutImpl2DNoneIndexStaticModule_basic",
Add aten.pool_max3d support to torch-to-linalg (#2735) Added verification logic to the abstract_interpreter_lib_gen.py Also made some unit tests Initially, I thought we can use `linalg::pooling_ndhwc_max` to help implement this problem. However, on a 5-dimensional matrix it does the pooling on dimensions (2, 3, 4) which is not what we want. We want pooling on dimensions (3, 4, 5). To achieve this, we would need to lower our code using the `linalg` dialect. Turns out the pooling code in `linalg` looks like this. ``` func @max_pooling_ncdhw(%I: memref<?x?x?x?x?xf32>, %K: memref<3xindex>, %O: memref<?x?x?x?x?xf32>, %strides: memref<3xindex>, %dilations: memref<3xindex>) { %c0 = arith.constant 0 : index %c1 = arith.constant 1 : index %N = memref.dim %I, %c0 : memref<?x?x?x?x?xf32> %C = memref.dim %I, %c1 : memref<?x?x?x?x?xf32> %D = memref.dim %I, 2 : memref<?x?x?x?x?xf32> %H = memref.dim %I, 3 : memref<?x?x?x?x?xf32> %W = memref.dim %I, 4 : memref<?x?x?x?x?xf32> %kernel_d = memref.load %K[%c0] : memref<3xindex> %kernel_h = memref.load %K[%c1] : memref<3xindex> %kernel_w = memref.load %K[2] : memref<3xindex> %stride_d = memref.load %strides[%c0] : memref<3xindex> %stride_h = memref.load %strides[%c1] : memref<3xindex> %stride_w = memref.load %strides[2] : memref<3xindex> %dilation_d = memref.load %dilations[%c0] : memref<3xindex> %dilation_h = memref.load %dilations[%c1] : memref<3xindex> %dilation_w = memref.load %dilations[2] : memref<3xindex> linalg.generic { indexing_maps = [ affine_map<(n, c, d, h, w, kd, kh, kw) -> (n, c, d * %stride_d + kd * %dilation_d, h * %stride_h + kh * %dilation_h, w * %stride_w + kw * %dilation_w)>, // Map for input tensor affine_map<(n, c, d, h, w, kd, kh, kw) -> (kd, kh, kw)>, // Map for kernel tensor affine_map<(n, c, d, h, w, kd, kh, kw) -> (n, c, d, h, w)> // Map for output tensor ], iterator_types = ["parallel", "parallel", "parallel", "parallel", "parallel", "reduction", "reduction", "reduction"], doc = "3D Max Pooling NCDHW with Strides, Dilations, and Kernel Size" } ins(%I, %K : memref<?x?x?x?x?xf32>, memref<3xindex>) outs(%O : memref<?x?x?x?x?xf32>) { ^bb0(%input_elem: f32, %kernel_elem: index, %output_elem: f32): %max_val = arith.maxf %input_elem, %output_elem : f32 linalg.yield %max_val : f32 } return } ``` This was implemented based on it's source code with the adjustments mentioned above: https://github.com/llvm/llvm-project/blob/4ca1b5e094280ef1af40412e3cfcb62dc3cf15bc/mlir/include/mlir/Dialect/Linalg/IR/LinalgNamedStructuredOps.yaml#L5647 Issues related to this can be found here https://github.com/nod-ai/SHARK-Turbine/issues/324
2024-01-19 23:39:46 +08:00
"MaxPool3dCeilModeTrueModule_basic",
"MaxPool3dEmptyStrideStaticModule_basic",
"MaxPool3dLargeDatadModule_basic",
"MaxPool3dModuleRandomSimple_basic",
"MaxPool3dModule_basic",
"MaxPool3dStaticCeilModeTrueModule_basic",
"MaxPool3dStaticModule_basic",
[TorchDynamo] Enable Elemtwise ops for Scalar arg (#2744) This commit provides dummy solution to support elmentwise operations (mul, add) with scalar argument. ( op(Tensor, Scalar) ) It replaces `torch.aten.add.Tensor` with `torch.aten.add.Scalar`. ``` Unexpected outcome summary: (torchdynamo) ****** Unexpectedly Passed tests - 22 tests XPASS - "AddCDivModule_basic" XPASS - "BatchNorm1DModule_basic" XPASS - "BatchNorm1DStaticShapeModule_basic" XPASS - "BatchNorm1DWith2DInputModule_basic" XPASS - "BatchNorm2DModule_basic" XPASS - "BatchNorm3DModule_basic" XPASS - "ElementwiseAddScalarInt64Module_basic" XPASS - "ElementwiseAddScalarIntModule_basic" XPASS - "ElementwiseMulScalarModule_basic" XPASS - "ElementwiseMulScalarModule_float" XPASS - "ElementwiseMulScalarModule_int" XPASS - "GroupNormModule_basic" XPASS - "GroupNormNoWeightAndBiasModule_basic" XPASS - "MobilenetV3Module_basic" XPASS - "NativeBatchNorm1DModule_basic" XPASS - "NativeBatchNorm2DModule_basic" XPASS - "NativeBatchNorm3DModule_basic" XPASS - "NativeBatchNormNoneWeightModule_basic" XPASS - "NativeGroupNormBackwardModule_basic" XPASS - "NativeGroupNormModule_basic" XPASS - "ResNet18Module_basic" XPASS - "ResNet18StaticModule_basic" ``` And segfault for test "ElementwiseAddScalar_TensorLiteralInt32_Module_basic". Somehow this change doesn't allow to use Tensors, that are not forward arguments, but local variables of model. e.g. `self.x = torch.tensor(..)` See also: #2745 Signed-off-by: Dmitrii Makarenko <dmitrii.makarenko@intel.com>
2024-03-12 03:22:05 +08:00
# Looks like incorrect fx graph conversion
"ElementwiseAddScalar_TensorLiteralInt32_Module_basic",
[torchdynamo] Initial TorchDynamo support This adds a basic e2e Config for TorchDynamo using Linalg-on-Tensors/RefBackend. But TorchDynamo is pretty orthogonal to various other pieces, so it should compose nicely with variations like: - Switching out all the backends (Linalg-on-Tensors, TOSA, MHLO) - PyTorch functionalization and decompositions - Taking the example inputs and compiling with all dynamic or all static shapes without duplicating tests. This adds it to the CI, but there are still a lot of XFAIL's. This also adds a helper `from torch_mlir.dynamo import make_simple_dynamo_backend` which simplifies some of the steps for making a Torch-MLIR-based TorchDynamo backend. We include "simple" in the name because we are going to be exploring various things next from the long-term roadmap. The next steps are: - Burn down all the XFAIL's. - Start working on the pieces from the [long-term roadmap](https://github.com/llvm/torch-mlir/blob/main/docs/long_term_roadmap.md). - Add functionalization/decompositions into the TorchDynamo flow and remove reliance on the current Torch-MLIR "frontend". - Write a pure-Python direct FX->MLIR importer. - Hook up the new PyTorch symbolic shape stuff. - Explore PrimTorch decompositions for simplifying backends.
2022-11-18 20:21:19 +08:00
}
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
FX_IMPORTER_XFAIL_SET = {
Add per-test timeouts to catch infinite loops (#3650) Previously we only had full suite timeouts, making it impossible to identify which specific tests were hanging. This patch adds: 1. Per-test timeout support in the test framework 2. A default 600s timeout for all tests 3. A deliberately slow test to verify the timeout mechanism works The timeout is implemented using Python's signal module. Tests that exceed their timeout are marked as failures with an appropriate error message. This should help catch and isolate problematic tests that enter infinite loops, without needing to re-run the entire suite multiple times.
2024-08-22 02:37:31 +08:00
"TimeOutModule_basic", # this test is expected to time out
[Stablehlo] support aten.any.dim, aten.min.dim (#3500) * refactor `TorchToStablehlo/Reduction.cpp` * add `ConvertAtenReduceWithIndicesOp` patterns
2024-06-29 16:53:33 +08:00
"ReduceAnyDimFloatModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
"ArangeStartOutViewModule_basic",
"AtenFloatScalarModule_basic",
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
[stablehlo] add aten::_int_mm lowering (#3474) as title
2024-06-20 16:10:31 +08:00
"AtenIntMM_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenItemFpOpModule_basic",
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
"AtenMatmulQint8_basic",
"AtenMatmulQint8VM_basic",
"AtenMatmulQint8VV_basic",
"AtenMmQMixedSigni8_basic",
"AtenMmQint8_basic",
"AtenMmQuint8_basic",
[torch] Adds Quantization Support for `aten.relu` (#3177) A choice was made to quantize the return type of Relu with a scale and zero point copied from the input's quantization scheme. With this choice, the torch-to-linalg conversion of quantized Relu essentially computes max(input, zeroPoint) in the elementwise payload.
2024-04-24 02:01:36 +08:00
"QuantizedReluInt32_basic",
"QuantizedReluInt8_basic",
"QuantizedReluUint8_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AtenSubFloatModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"BroadcastDynamicDimModule_basic",
"CeilFloatModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ConvTbcModule_basic",
[TorchToLinalg] Adds Quantization Support for ConvTranspose (#3240) I spent a little while debugging numerics issues with some tests similar to the ones in quantized_models.py, only to find that pytorch's quantized conv transpose is catastrophically inaccurate. I'll upstream the issue and only leave the tests here which are of the form quantize -> dequantize -> op.
2024-05-01 00:23:09 +08:00
"ConvTranspose2DQInt8_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2DStrided_basic",
"ConvolutionBackwardModule2D_basic",
"CumsumModule_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
"DeformConv2D_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"DivFloatModule_basic",
"DivIntModule_basic",
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
"ElementwiseDequantizePerChannelModule_basic",
"ElementwiseDequantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorUIntModule_basic",
[Torch] Emit rrelu and decompose it (#3250) as title
2024-06-03 19:25:52 +08:00
"ElementwiseRreluEvalModule_basic",
"ElementwiseRreluEvalStaticModule_basic",
"ElementwiseRreluTrainModule_basic",
"ElementwiseRreluTrainStaticModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ElementwiseToDtypeI64ToUI8Module_basic",
"EqIntModule_basic",
"FloatImplicitModule_basic",
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
"GtFloatIntModule_basic",
"GtIntModule_basic",
"IntFloatModule_basic",
"IntImplicitModule_basic",
"LenStrModule_basic",
"MulFloatModule_basic",
"NativeGroupNormBackwardModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"PowIntFloatModule_basic",
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
"QuantizedBatchedInputSingleLayer_basic",
"QuantizedMLP_basic",
"QuantizedNoLayer_basic",
"QuantizedSingleLayer_basic",
"ReduceMaxAlongDimUnsignedInt_basic",
"ReduceMinAlongDimUnsignedInt_basic",
"RsubInt0d_NumToTensor_Module_basic",
"ScalarImplicitFloatModule_basic",
implement lowering of torch.aten._linalg_slogdet (#3524)
2024-07-19 13:54:43 +08:00
"SignAndLogarithmOfDeterminantModule_F32",
"SignAndLogarithmOfDeterminantBatchedModule_F32",
"SignAndLogarithmOfDeterminantDynamicModule_F32",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"SortIntListReverse_basic",
"SortIntList_basic",
"SplitDimDynamicModule_basic",
"SplitDimStaticModule_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SubFloatModule_basic",
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"ThresholdBackward2dMixedModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"UpSampleNearest2dDynamicFactor_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewSizeFromOtherTensor_basic",
[Torch] Implements TorchToLinalg lowering of torch.ops.aten._weight_norm_interface (#3538) Resolves https://github.com/nod-ai/SHARK-Turbine/issues/757. Adds TorchToLinalg lowering for `Aten_WeightNormInterfaceOp`. --------- Co-authored-by: Ubuntu <rbhowmik@RohanBhowmikVM.judsoscro3wupi0qm4bjlj5m3b.bx.internal.cloudapp.net>
2024-07-17 01:39:12 +08:00
"WeightNormInterfaceModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
# REMOVE WHEN ENABLE_GQA IS ADDED
"ScaledDotProductAttentionBoolMaskModule_basic",
"ScaledDotProductAttentionDifferentCausalModule_basic",
build: manually update PyTorch version (#3666) Set PyTorch and TorchVision version to nightly release 2024-08-25. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-28 00:28:30 +08:00
"ScaledDotProductAttentionDifferentModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
"ScaledDotProductAttentionMaskModule_basic",
"ScaledDotProductAttentionSameCausalModule_basic",
"ScaledDotProductAttentionSameDynamicModule_basic",
build: manually update PyTorch version (#3666) Set PyTorch and TorchVision version to nightly release 2024-08-25. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-28 00:28:30 +08:00
"ScaledDotProductAttentionSameModule_basic",
[FxImporter] Fix fx importer test config and clean xfail set (#3176)
2024-04-17 13:36:07 +08:00
}
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
FX_IMPORTER_CRASHING_SET = LINALG_CRASHING_SET | {
[FxImporter] Fix fx importer test config and clean xfail set (#3176)
2024-04-17 13:36:07 +08:00
"HBC_basic",
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
# Runtime op verification: out-of-bounds access
"_SoftmaxModule_basic",
"UpSampleNearest2dDynamicFactor_basic",
[FxImporter] Add FxImporter config in e2e-test (#3151)
2024-04-13 07:07:56 +08:00
}
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
FX_IMPORTER_STABLEHLO_XFAIL_SET = {
[ci] enable fx_importer2stablehlo ci test (#3698)
2024-09-11 09:53:23 +08:00
"AddFloatIntModule_basic",
"ArgmaxIntModule_basic",
"ArgmaxIntModule_multiple_maxs",
"ArgmaxKeepdimModule_basic",
"ArgmaxModule_basic",
"AtenKthvalueDynamicDimsModule_basic",
"AtenKthvalueFloat64DynamicDimsModule_basic",
"AtenKthvalueFloat64Module_basic",
"AtenKthvalueKeepDimModule_basic",
"AtenKthvalueModule_basic",
"AtenPolarDoubleModule_basic",
"AtenPolarFloatModule_basic",
"DiagonalWithStaticShapeModule_basic",
"EinsumStaticDiagonalDimensionModule_basic",
"ElementwiseIntTensorLtFloatScalarModule_basic",
"ElementwiseRemainderScalarModule_Bool_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Float_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Float_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Float_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDivisor_basic",
"ElementwiseRreluEvalModule_basic",
"ElementwiseRreluEvalStaticModule_basic",
"ElementwiseRreluTrainModule_basic",
"ElementwiseRreluTrainStaticModule_basic",
"MaxPool1dCeilModeTrueModule_basic",
"MaxPool1dStaticCeilModeTrueModule_basic",
"MaxUnpool3dModulePad0_basic",
"MaxUnpool3dModule_basic",
"MultinomialModule2D_F32",
"MultinomialModule2D_basic",
"MultinomialModule_basic",
"QuantizedReluInt32_basic",
"QuantizedReluInt8_basic",
"QuantizedReluUint8_basic",
"ScatterAddStaticModule_basic",
"SelectScattertModule_basic",
"SelectScattertStaticModule_basic",
"SignAndLogarithmOfDeterminantBatchedModule_F32",
"SignAndLogarithmOfDeterminantDynamicModule_F32",
"SignAndLogarithmOfDeterminantModule_F32",
"SliceCopyEndGreaterThanDimSize_Module_basic",
"SliceCopyNegative_Module_basic",
"SliceCopyNonZeroDim_Module_basic",
"SliceCopyStartGreaterThanDimSize_Module_basic",
"SliceCopy_Module_basic",
"SliceScatterModule_basic",
"SliceScatterNegativeDimModule_basic",
"SliceScatterNegativeEndModule_basic",
"SliceScatterStaticModule_basic",
"SliceScatterStepVariationModule_basic",
"SliceScatterZeroDimModule_basic",
"TimeOutModule_basic",
"WeightNormInterfaceModule_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"AdaptiveAvgPool3dDynamicNoBatch_basic",
"AdaptiveAvgPool3dDynamic_basic",
"AdaptiveMaxPool1dDynamicNoBatch_basic",
"AdaptiveMaxPool1dDynamic_basic",
"AdaptiveMaxPool1dStatic_basic",
"AdaptiveMaxPool2dDynamicNoBatch_basic",
"AdaptiveMaxPool2dDynamicWithIndices_basic",
"AdaptiveMaxPool2dDynamic_basic",
"AdaptiveMaxPool2dStaticWithIndices_basic",
"AdaptiveMaxPool2dStatic_basic",
"AdaptiveMaxPool3dDynamicNoBatch_basic",
"AdaptiveMaxPool3dDynamicWithIndices_basic",
"AdaptiveMaxPool3dDynamic_basic",
"AdaptiveMaxPool3dStaticWithIndices_basic",
"AdaptiveMaxPool3dStatic_basic",
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
"ArangeStartOutViewModule_basic",
"ArgminIntModule_basic",
"ArgminIntModule_multiple_mins",
"ArgminModule_basic",
"AtenComplexImagModule_basic",
"AtenComplexRealModule_basic",
"AtenComplexViewModule_basic",
"AtenDiagEmbedDefaultDiag_basic",
"AtenDiagEmbedDimDiag_basic",
"AtenDiagEmbedNegOffsetDiag_basic",
"AtenDiagEmbedNonDefault4DDiag_basic",
"AtenDiagEmbedOffsetDiag_basic",
"AtenDiagEmbedRevDimDiag_basic",
"AtenEmbeddingBagSumExample_basic",
"AtenFloatScalarModule_basic",
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
"AtenItemFpOpModule_basic",
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
"AtenMatmulQint8VM_basic",
"AtenMatmulQint8VV_basic",
"AtenMatmulQint8_basic",
"AtenMmQMixedSigni8_basic",
"AtenMmQint8_basic",
"AtenMmQuint8_basic",
"AtenRealView128Module_basic",
"AtenRealView64Module_basic",
"AtenSubFloatModule_basic",
"AtenTopKModule_basic",
"AtenTopKSmallestModule_basic",
"Aten_EmbeddingBagExample_basic",
"AvgPool2dDivisorOverrideModule_basic",
"BernoulliTensorModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"BroadcastDynamicDimModule_basic",
"CeilFloatModule_basic",
"ConstantBoolParameterModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"ConvTbcModule_basic",
[TorchToLinalg] Adds Quantization Support for ConvTranspose (#3240) I spent a little while debugging numerics issues with some tests similar to the ones in quantized_models.py, only to find that pytorch's quantized conv transpose is catastrophically inaccurate. I'll upstream the issue and only leave the tests here which are of the form quantize -> dequantize -> op.
2024-05-01 00:23:09 +08:00
"ConvTranspose2DQInt8_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2DStrided_basic",
"ConvolutionBackwardModule2D_basic",
"CumsumModule_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
"DeformConv2D_basic",
[TorchToLinalg][ONNX] Add Basic Determinant Support (#3481) This adds support for a few ops: - torch.linalg_det - torch._linalg_det (if the LU and pivot returns are unused) - onnx.Det An scf loop is used, since the row reduction algorithm applied here has some loop-carried dependencies. The current support being added here is very basic, and only works if no permutations are required during row reduction, and assumes the matrices are non-singular.
2024-06-26 02:34:19 +08:00
"DeterminantBatchedModule_F32",
"DeterminantDynamicModule_F32",
"DeterminantModule_F32",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"DiagonalModule_basic",
"DiagonalModule_nonsquare",
"DiagonalModule_transposed",
"DiagonalModule_with_dims",
"DiagonalModule_with_dims_and_offset",
"DiagonalModule_with_negative_dims",
"DiagonalModule_with_offset",
"DivFloatModule_basic",
"DivIntModule_basic",
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
"ElementwiseDequantizePerChannelModule_basic",
"ElementwiseDequantizePerTensorModule_basic",
"ElementwiseErfIntModule_basic",
"ElementwiseLogitModule_basic",
"ElementwiseMulTensorComplexModule_basic",
[linalg] Fix bug for conversion of complex dtype (#3269) The conversion of complex type wasn't supported or checked; the support and required tests were added. Fixes: https://github.com/iree-org/iree/issues/17226#issuecomment-2087779158
2024-05-01 14:36:53 +08:00
"ElementwiseMulTensorComplexDiffModule_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"ElementwiseQuantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorUIntModule_basic",
"ElementwiseReciprocalIntModule_basic",
"ElementwiseToDtypeI64ToUI8Module_basic",
"EqIntModule_basic",
"Fill_TensorFloat32WithFloat32_basic",
"Fill_TensorFloat32WithFloat64_basic",
"Fill_TensorFloat32WithInt64_basic",
"FloatImplicitModule_basic",
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
"GtFloatIntModule_basic",
"GtIntModule_basic",
"HBC_basic",
"HardtanhBackward_basic",
"IndexPut1DFloatAccumulateModule_basic",
"IndexPut1DFloatNonAccumulateModule_basic",
"IndexPut1DIntAccumulateModule_basic",
"IndexPut1DIntNonAccumulateModule_basic",
"IndexPut2DFloatAccumulateModule_basic",
"IndexPut2DFloatNonAccumulateModule_basic",
"IndexPut2DIntAccumulateModule_basic",
"IndexPut2DIntNonAccumulateModule_basic",
"IndexPut3DFloatAccumulateModule_basic",
"IndexPut3DFloatNonAccumulateModule_basic",
"IndexPut3DIntAccumulateModule_basic",
"IndexPut3DIntNonAccumulateModule_basic",
"IndexPutHackedTwin1DFloatAccumulateModule_basic",
"IndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin1DIntAccumulateModule_basic",
"IndexPutHackedTwin1DIntNonAccumulateModule_basic",
"IndexPutHackedTwin2DFloatAccumulateModule_basic",
"IndexPutHackedTwin2DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin2DIntAccumulateModule_basic",
"IndexPutHackedTwin2DIntNonAccumulateModule_basic",
"IndexPutHackedTwin3DFloatAccumulateModule_basic",
"IndexPutHackedTwin3DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin3DIntAccumulateModule_basic",
"IndexPutHackedTwin3DIntNonAccumulateModule_basic",
"IndexPutImpl1DFloatAccumulateModule_basic",
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
"IndexPutImpl2DFloatAccumulateModule_basic",
"IndexPutImpl2DFloatNonAccumulateModule_basic",
"IndexPutImpl2DImplicitModule_basic",
"IndexPutImpl2DIndexModule_basic",
"IndexPutImpl2DNoneIndexStaticModule_basic",
"IndexPutImpl3DFloatAccumulateModule_basic",
"IndexPutImpl3DFloatNonAccumulateModule_basic",
"IndexPutImplIndexWithNoneModule_basic",
[StableHLO] Fix aten.clamp.Tensor in FxImporter2StableHLO (#3190) The FX importer will pass static shapes to the Torch dialect, so it needs to generate a StableHLO that satisfies shape inference.
2024-04-19 17:08:29 +08:00
"IndexSelectRank0IdxModule_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"IndexTensorNegativeIndexModule_basic",
"IntFloatModule_basic",
"IntImplicitModule_basic",
"LenStrModule_basic",
"MaxPool2dCeilModeTrueModule_basic",
"MaxPool2dStaticCeilModeTrueModule_basic",
"MaxPool2dWithIndicesBackwardDynamic3DModule_basic",
"MaxPool2dWithIndicesBackwardDynamic4DModule_basic",
"MaxPool2dWithIndicesBackwardStatic3DModule_basic",
"MaxPool2dWithIndicesBackwardStatic4DModule_basic",
"MaxPool3dCeilModeTrueModule_basic",
"MaxPool3dEmptyStrideStaticModule_basic",
"MaxPool3dLargeDatadModule_basic",
"MaxPool3dModuleRandomSimple_basic",
"MaxPool3dModule_basic",
"MaxPool3dStaticCeilModeTrueModule_basic",
"MaxPool3dStaticModule_basic",
[TorchToLinalg] Support lowering MaxPool3dWithIndices (#3652) Support torch.MaxPool3dWithIndices lowering to linalg backend.
2024-08-28 03:14:25 +08:00
"MaxPool3dWithIndicesAllNegativeValuesModule_basic",
"MaxPool3dWithIndicesAllOnesModule_basic",
"MaxPool3dWithIndicesCeilModeTrueModule_basic",
"MaxPool3dWithIndicesFullSizeKernelModule_basic",
"MaxPool3dWithIndicesModule_basic",
"MaxPool3dWithIndicesNonDefaultDilationModule_basic",
"MaxPool3dWithIndicesNonDefaultPaddingModule_basic",
"MaxPool3dWithIndicesNonDefaultParamsModule_basic",
"MaxPool3dWithIndicesNonDefaultStrideModule_basic",
"MaxPool3dWithIndicesStaticModule_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"MulFloatModule_basic",
"NativeGroupNormBackwardModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
"NllLossModuleBackwardMeanWeight_basic",
"NllLossModuleBackwardMean_basic",
"NllLossModuleBackwardSumWeight_basic",
"NllLossModuleBackwardSum_basic",
"NllLossModuleBackwardWeight_basic",
"NllLossModuleBackward_basic",
"NllLossModuleBackward_ignore_index",
"NormScalarComplexModule_basic",
"NormScalarModule_basic",
"NormalFunctionalModule_basic",
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"PowIntFloatModule_basic",
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
"QuantizedBatchedInputSingleLayer_basic",
"QuantizedMLP_basic",
"QuantizedNoLayer_basic",
"QuantizedSingleLayer_basic",
"RandnDtypeDeviceModule_basic",
"RandnGeneratorF64Module_basic",
"RandnGeneratorModule_basic",
"RandnLikeDtypeModule_basic",
"RandnLikeModule_basic",
"RandnModule_basic",
"ReduceAllDimBool_basic",
"ReduceAllDimEmpty_basic",
"ReduceAllDimFloat_basic",
"ReduceAllDimInt_basic",
"ReduceProdDimIntFloatModule_basic",
"ReflectionPad1dModule2dInput_Right",
"ReflectionPad1dModule2dInput_basic",
"ReflectionPad1dModule3dInput_Left",
"ReflectionPad1dModule3dInput_basic",
"ReflectionPad2dModule_Bottom",
"ReflectionPad2dModule_Left",
"ReflectionPad2dModule_Right",
"ReflectionPad2dModule_Top",
"ReflectionPad2dModule_basic",
"ReplicationPad2dModule_basic",
"ReplicationPad2dModule_bottom0",
"ReplicationPad2dModule_left0",
"ReplicationPad2dModule_right0",
"ReplicationPad2dModule_top0",
"RsubInt0d_NumToTensor_Module_basic",
"ScalarImplicitFloatModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
# REMOVE WHEN ENABLE_GQA IS ADDED
"ScaledDotProductAttentionBoolMaskModule_basic",
"ScaledDotProductAttentionDifferentCausalModule_basic",
"ScaledDotProductAttentionDifferentModule_basic",
"ScaledDotProductAttentionMaskModule_basic",
"ScaledDotProductAttentionSameCausalModule_basic",
"ScaledDotProductAttentionSameDynamicModule_basic",
"ScaledDotProductAttentionSameModule_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
"ScatterReduceFloatMaxModule",
"ScatterReduceFloatMaxModuleIncludeSelf",
"ScatterReduceFloatMeanModule",
"ScatterReduceFloatMeanModuleIncludeSelf",
"ScatterReduceFloatMinModule",
"ScatterReduceFloatMinModuleIncludeSelf",
"ScatterReduceFloatProdModule",
"ScatterReduceFloatProdModuleIncludeSelf",
"ScatterReduceFloatSumModule",
"ScatterReduceFloatSumModuleIncludeSelf",
"ScatterReduceIntMaxModule",
"ScatterReduceIntMaxModuleIncludeSelf",
"ScatterReduceIntMeanModule",
"ScatterReduceIntMeanModuleIncludeSelf",
"ScatterReduceIntMinModule",
"ScatterReduceIntMinModuleIncludeSelf",
"ScatterReduceIntProdModule",
"ScatterReduceIntProdModuleIncludeSelf",
"ScatterReduceIntSumModule",
"ScatterReduceIntSumModuleIncludeSelf",
"ScatterSrcModule_basic",
"ScatterSrcStaticModule_basic",
"ScatterValueFloatModule_basic",
"ScatterValueIntModule_basic",
"SliceOutOfLowerBoundEndIndexModule_basic",
"SortIntListReverse_basic",
"SortIntList_basic",
"SortTensorDescending_basic",
"SortTensorInteger_basic",
"SortTensorNegativeDimension_basic",
"SortTensorSpecificDimension_basic",
"SortTensor_basic",
"SplitDimDynamicModule_basic",
"SplitDimStaticModule_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SubFloatModule_basic",
"TModuleRank0_basic",
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"TensorToInt_basic",
"Threshold1dFloatModule_basic",
"Threshold1dIntI32Module_basic",
"Threshold1dIntModule_basic",
"Threshold2dFloatModule_basic",
"Threshold2dIntModule_basic",
"Threshold3dFloatModule_basic",
"Threshold3dIntModule_basic",
"ThresholdBackward1dFloatModule_basic",
"ThresholdBackward1dIntModule_basic",
"ThresholdBackward1dMixedModule_basic",
"ThresholdBackward2dFloatModule_basic",
"ThresholdBackward2dIntModule_basic",
"ThresholdBackward2dMixedModule_basic",
"ThresholdBackward3dFloatModule_basic",
"ThresholdBackward3dIntModule_basic",
"ThresholdBackward3dMixedModule_basic",
"TraceModule_basic",
"TraceModule_empty",
"TraceModule_nonsquare",
"TraceSignedIntModule_basic",
"TraceUnsignedIntModule_basic",
"TraceUnsignedIntModule_empty",
"UnsafeIndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"UpSampleNearest2dBackwardScalesNone_basic",
"UpSampleNearest2dBackward_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewSizeFromOtherTensor_basic",
}
FX_IMPORTER_STABLEHLO_CRASHING_SET = {
"BatchNorm1DModule_basic",
"BatchNorm2DModule_basic",
"BatchNorm3DModule_basic",
"ResNet18Module_basic",
"ResNet18StaticModule_basic",
"MobilenetV3Module_basic",
"Conv2dBiasNoPaddingModule_basic",
[ci] enable fx_importer2stablehlo ci test (#3698)
2024-09-11 09:53:23 +08:00
# llvm-project/llvm/include/llvm/ADT/ArrayRef.h:257:
# const T &llvm::ArrayRef<long>::operator[](size_t) const [T = long]:
# Assertion `Index < Length && "Invalid index!"
"IndexPutWithNoneAndBroadcastModule_basic",
# Assertion `newMaterialization.getType() == outputType
# materialization callback produced value of incorrect type failed
"ReduceMaxAlongDimUnsignedInt_basic",
"ReduceMinAlongDimUnsignedInt_basic",
[FxImporter] Add fx importer to stablehlo e2e test config (#3183)
2024-04-19 12:29:17 +08:00
}
mhlo: migrate conversion to stablehlo (#1840) This patch replaces all MHLO operations with their StableHLO counterparts and adds a validation pass to ensure that no MHLO operations remain before translating all Stablehlo operations to the MHLO dialect for further lowering to the Linalg dialect. This patch also updates all lit tests so that they refer to the `convert-torch-to-stablehlo` pass and so that they check for StableHLO operations.
2023-02-02 21:29:47 +08:00
STABLEHLO_PASS_SET = {
[Torch] support decomposition of aten.aminmax (#3513) * unify decompisition of `aten.amax` and `aten.amin` * support `aten.amax` with `dim=()`
2024-06-29 21:44:05 +08:00
"ReduceAminmaxSingleDim_basic",
"ReduceAminmaxAllDims_basic",
"ReduceAmaxEmptyDim_basic",
[Stablehlo] support aten.any.dim, aten.min.dim (#3500) * refactor `TorchToStablehlo/Reduction.cpp` * add `ConvertAtenReduceWithIndicesOp` patterns
2024-06-29 16:53:33 +08:00
"ReduceMinAlongDimNegative_basic",
"ReduceMinAlongDim_basic",
"ArgminModule_with_dim",
"ReduceMinAlongDimSignedInt_basic",
"ReduceAnyDimFloatModule_basic",
[Torch] Add support for Meshgrid (#3462)
2024-06-14 23:59:08 +08:00
"MeshgridIndexingIJ_basic",
"MeshgridIndexingXY_basic",
"Meshgrid_basic",
[Torch] support recompose of aten.split.with_sizes and aten.tensor_sp… (#3401) …lit.sections * support recompose to aten.split.with_sizes and aten.tensor_split.sections * fix recompose of aten.chunk
2024-05-31 09:56:47 +08:00
"SplitWithSizes_Module_basic",
"TensorSplitSections_GetItemModule_basic",
"TensorSplitSections_ListUnpackModule_basic",
[Stablehlo] support uint8 (#3367) Support lowering unsigned integer type to stablehlo as discussed in https://github.com/llvm/torch-mlir/pull/2184. The things I do in this PR: 1. create `setupBackendTypeConversionForStablehlo()`, `createFuncBackendTypeConversionForStablehloPass` and `createFinalizingBackendTypeConversionForStablehloPass`. 2. remove `InferTypeOpInterface` from `torch_c.to_builtin_tensor`, because it's different result type between linalg backend and stablehlo backend: ``` // linalg backend func.func @forward(%arg0: !torch.vtensor<[3],ui8>) -> tensor<3xf32> { %c = torch_c.to_builtin_tensor %arg0 : (!torch.vtensor<[3], ui8> -> tensor<3xi8> %0 = tensor.empty() : tensor<3xf32> %1 = linalg.generic {indexing_maps = [#map, #map], iterator_types = ["parallel"]} ins(%arg0 : tensor<3xi8>) outs(%0 : tensor<3xf32>) { ^bb0(%in: i8, %out: f32): %2 = arith.uitofp %in : i8 to f32 linalg.yield %2 : f32 } -> tensor<3xf32> return %1 : tensor<3xf32> } // stablehlo backend func.func @forward(%arg0: !torch.vtensor<[3],ui8>) -> tensor<3xf32> { %c = torch_c.to_builtin_tensor %arg0 : (!torch.vtensor<[3], ui8> -> tensor<3xui8> %0 = stablehlo.convert %arg0 : (tensor<3xui8> -> tensor<3xf32> return %0 : tensor<3xf32> } ``` 3. fix stablehlo and linalg's conversion
2024-06-04 09:04:59 +08:00
"EmptyModule_uint8",
"TypeConversionUint8ToF32Module_basic",
Implement lowering of torch.aten.atleast_1d (#3498) This operator is necessary in order to implement torch.aten.vstack. Which will be added in a future PR.
2024-07-17 20:50:30 +08:00
"Atleast1dModule0dInput_basic",
"Atleast1dModule1dInput_basic",
Implement lowering of aten.atleast_2d (#3546) This operator is needed to implement aten.vstack, which will be submitted in a subsequent PR
2024-08-14 21:22:31 +08:00
"Atleast2dModule0dInput_basic",
"Atleast2dModule1dInput_basic",
"Atleast2dModule2dInput_basic",
[Torch] fix aten.linear's decomposition (#3391) * support aten.linear with more rank.
2024-05-27 15:49:50 +08:00
"AtenLinear1D_basic",
"AtenLinear2D_basic",
"AtenLinear3DBias_basic",
"AtenLinearMatVec_basic",
"AtenLinearVecMatBias_basic",
"AtenLinearVecMat_basic",
[Stablehlo] refactor reduction lowering and support aten.amin (#3383) * implement detailed lowering template pattern `ConvertAtenReduceAllDimsOp` and `ConvertAtenReduceKeepDimOp` * support `aten.amin`'s lowering.
2024-05-23 20:40:20 +08:00
"ReduceAminSingleDim_basic",
[Torch] emit aten.dot and canonicalize it to aten.matmul (#3361) * canonicalize `aten.dot` to `aten.matmul`
2024-05-18 22:45:14 +08:00
"AtenDotModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"AdaptiveAvgPool1dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool1dUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool2dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool2dUnitOutputSizeStaticModule_basic",
[Torch] Add general AdaptiveAvgPool2dOp decompose support (#3111) Previously, it could only handle the situations where outputsize == (1, 1) or outputsize == (input_H, input_W). Now it supports all situations where input_H % output_H== 0 && input_W % output_W == 0
2024-04-11 17:02:59 +08:00
"AdaptiveAvgPool2dOutputSizeDivisibleByInputStaticModule_basic",
[Torch] Add support for static uneven divisible AdaptiveAvgPool2d (#3566) The static uneven divisible AdaptiveAvgPool2d means that although the input size is not an integer multiple of ouput size, but the kernel and stride size can also be fixed (not dynamic). The derivation logic of kernel and stride size is consistent with torch/_decomp/decomposations.py:adaptive_avg_pool2d as described in the following: 1. Stride Size Firstly , derive the start index in each reduce operation according to the output size (`n`), `start_index = ([0, 1, ..., n - 1] * input_size) // output_size`. For each index `k`, if `k * (input_size % output_size) < output_size`, then the current and previous stride keeps the same as `input_size // output_size`. So suppose `(n-1) * (input_size % output_size) < output_size`, the stride in the whole AdaptiveAvgPool2d process keeps static, as `input_size // output_size`. 2. Kernel Size torch/_decomp/decomposations.py:adaptive_avg_pool2d calculates a static kernel size when the input/output sizes satisfy either of the two conditions, `input_size % output_size == 0` or `output_size % (input_size % output_size) == 0`. Here if `input_size % output_size == 0`, then the kernel size equals `input_size // output_size`, otherwise `input_size // output_size + 1.`
2024-08-01 11:37:53 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
[Stablehlo] Add converter to stablehlo for aten.(Int,Float,Bool).Tensor op (#2340) [Stablehlo] Add converter to stablehlo for aten.(Int,Float,Bool).Tensor op and configure crashing e2e sets for stablehlo backend.
2023-07-29 21:55:49 +08:00
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[MLIR][TORCH] Add e2e support for aten.alias -- This commit adds e2e support for aten.alias op. Signed-off-by: Abhishek Varma <abhishek@nod-labs.com>
2023-06-21 01:14:09 +08:00
"AliasModule_basic",
[Torch] Add support for Aten__Or__BoolOp (#3574)
2024-07-31 17:23:53 +08:00
"TrueFalseOrBoolOpModule_basic",
[Stablehlo] Enable Stablehlo backend with arith dialect (#2139)
2023-05-26 22:57:57 +08:00
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
Add torch-to-mhlo e2e support for AtenArangeStartStepOp (#1385) Co-authored-by: Vremold <xremold@gamil.com>
2022-09-20 22:31:24 +08:00
"ArangeDtypeFloatModule_basic",
"ArangeDtypeIntModule_basic",
"ArangeFalsePinMemoryModule_basic",
"ArangeFloatModule_basic",
"ArangeIntModule_basic",
"ArangeNegativeStartFloatModule_basic",
"ArangeNegativeStartIntModule_basic",
"ArangeStartFloatModule_basic",
"ArangeStartIntModule_basic",
"ArangeStartNegativeStepFloatModule_basic",
"ArangeStartNegativeStepIntModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ArangeStartOutDtypeModule_basic",
"ArangeStartOutModule_basic",
Add torch-to-mhlo e2e support for AtenArangeStartStepOp (#1385) Co-authored-by: Vremold <xremold@gamil.com>
2022-09-20 22:31:24 +08:00
"ArangeStartStepFloatModule_basic",
"ArangeStartStepIntModule_basic",
"ArangeZeroElementOutputModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ArgmaxModule_with_dim",
"AtenComplex64Module_basic",
"AtenFloatScalarModule_basic",
Implement lowering for torch.aten.hann_window.periodic (#3502)
2024-07-17 20:51:23 +08:00
"AtenHannWindowPeriodicFalseModule_basic",
"AtenHannWindowPeriodicTrueModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
"AtenIntTensorByteDtypeModule_basic",
"AtenIntTensorCharDtypeModule_basic",
"AtenItemIntOpModule_basic",
"AtenMmFloatTypes_basic",
"AtenMmIntTypes_basic",
[stablehlo] add aten::_int_mm lowering (#3474) as title
2024-06-20 16:10:31 +08:00
"AtenIntMM_basic",
[Stablehlo] lowering aten.round to stablehlo.round_nearest_even (#3011)
2024-03-12 08:58:20 +08:00
"AtenRoundFloatHalfToEvenModule_basic",
"AtenRoundFloatModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"AtenRoundIntModule_basic",
"AtenSubFloatModule_basic",
"AtenToDeviceModule_basic",
[Stablehlo] Support AtenTrilOp (#3359) 1. lower aten.tril to stablehlo composed by iota, select and so forth 2. add related e2e test cases
2024-05-20 15:49:24 +08:00
"AtenTrilStaticModule_basic",
"AtenTrilWithNegDiagonalStaticModule_basic",
"AtenTrilWithPosDiagonalStaticModule_basic",
[Torch] Support Aten_CastFloatOp. (#3115) By canonicalize Aten_CastFloatOp into AtenToDtypeOp
2024-04-09 11:06:53 +08:00
"Aten_CastFloatModule_basic",
[Torch] Support Aten_CastLongOp. (#3160) By canonicalize Aten_CastLongOp into AtenToDtypeOp
2024-04-17 21:58:32 +08:00
"Aten_CastLongModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"AvgPool1dStaticModule_basic",
"AvgPool2dStaticModule_basic",
[ONNX] Fix AveragePool attributes support (#3235) Issues was found here https://github.com/nod-ai/SHARK-Turbine/issues/643 - [ONNX] Fix padding attributes for onnx.AveragePool - [Linalg] Add countIncludePad false support for AtenAvgPool1/2dOp - [Linalg] Add an avg_pool2d countIncludePad False e2e tests - [Linalg] Fix conflict with AtenAvgPool3dOp - [Linalg] Fix e2e crash with AtenAvgPool1dOp - [Linalg] Add dynamic dim support for AtenAvgPool2dOp - [Linalg] Fix AvgPool2dDivisorOverrideModule crash
2024-06-13 03:16:43 +08:00
"AvgPool2dCountIncludePadFalseStaticModule_basic",
[MLIR][Torch] Add TorchToLinalg lowering for AtenAvgPool3dOp (#3030) This commit also fixes the average pool op' test failing for OnnxToLinalg lowering. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-06-05 00:42:34 +08:00
"AvgPool3dStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"BaddbmmBroadcast1DInputModule_basic",
"BaddbmmBroadcast2DInputModule_basic",
"BaddbmmStaticModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"BoolTensorReturnFalseModule_basic",
"BoolTensorReturnMixedModule_basic",
"BoolTensorReturnTrueModule_basic",
"BroadcastListConstructWithMinusOneModule_basic",
fix mhlo e2e ci crashes (#1620) * fix mhlo e2e ci crashes * add passed tests * calc dynamic positive dim
2022-11-21 21:50:35 +08:00
"BroadcastToSameRankStaticModule_basic",
"BroadcastZeroRankInputStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"CeilFloatModule_basic",
"ChunkListUnpackUneven_Module_basic",
"ChunkListUnpack_Module_basic",
"CloneModule_basic",
[Torch] Add decomposition of RepeatInterleaveSelfInt Op (#3075) Decomposition RepeatInterleaveSelfInt with following ops: ```python def my_repeat_interleave(input, repeats, dim=None): if dim is None: # Flatten the input and then repeat return input.flatten().unsqueeze(-1).tile((1, repeats)).flatten() else: # Calculate the shape after repeat expanded_shape = list(input.shape) expanded_shape[dim] *= repeats # Repeat the tensor along the specified dimension repeat_shape = [1] * (input.dim() + 1) repeat_shape[dim + 1] = repeats input = input.unsqueeze(-1) # Tile and then reshape tiled = torch.tile(input, repeat_shape) # Rearrange and reshape repeated = tiled.reshape(*expanded_shape) return repeated ``` I passed the tests of stablehlo and linalg. When testing onnx, strange things happened. In torch-mlir's CI **torch_nightly** and my own environment(torch==2.4.0.dev20240318+cpu), it can **pass the pass**. In torch-mlir's CI **torch_stable**, it **failed**. The test case is `RepeatInterleaveSelfIntNoDimModule_basic`, the result shape should be [120]. ```python class RepeatInterleaveSelfIntNoDimModule(torch.nn.Module): def __init__(self): super().__init__() @export @annotate_args([ None, ([3, 4, 5], torch.float32, True), ]) def forward(self, x): return x.repeat_interleave(2) @register_test_case(module_factory=lambda: RepeatInterleaveSelfIntNoDimModule()) def RepeatInterleaveSelfIntNoDimModule_basic(module, tu: TestUtils): module.forward(tu.rand(3, 4, 5)) ``` The error log is as follows: ``` Unexpected outcome summary: (onnx) ****** Failed tests - 1 tests FAIL - "RepeatInterleaveSelfIntNoDimModule_basic" @ trace item #0 - call to "forward" @ output of call to "forward" ERROR: shape (torch.Size([6, 4, 5])) is not equal to golden shape (torch.Size([120])) ``` @rsuderman Would you please help me check what's wrong with my PR? Thanks a lot.
2024-04-18 06:27:51 +08:00
"CollapseAllDimensionsModule_basic",
"CollapseStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ConstantBoolParameterModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"ContiguousModule_basic",
"Conv2dWithPaddingDilationStrideStaticModule_basic",
"Conv2dWithPaddingDilationStrideStaticModule_depthwise",
"Conv2dWithPaddingDilationStrideStaticModule_depthwise_multiplier",
"Conv2dWithPaddingDilationStrideStaticModule_grouped",
"Conv2dWithPaddingDilationStrideStaticModule_grouped_multiplier",
"Convolution2DStaticModule_basic",
"ConvolutionBackwardModule2DStatic_basic",
"ConvolutionModule2DTransposeStridedStatic_basic",
[Torch]Support conv_transpose1d and conv_transpose3d (#3286) 1. Support conv_transpose1d and conv_transpose3d 2. Fix bugs of convertTransposedConv func in lib/Conversion/TorchToStablehlo/Linear.cpp
2024-06-03 15:11:12 +08:00
"Conv_Transpose1dStaticModule_basic",
"Conv_Transpose2dStaticModule_basic",
"Conv_Transpose3dStaticModule_basic",
[stablehlo] Support ConstantPadNdOp in stablehlo (#3211) as title
2024-04-24 14:15:11 +08:00
"ConstantPad2dStaticModule_basic",
"ConstantPadNdModule_basic",
"ConstantPadNdPartialStaticModule_basic",
"ConstantPadNdStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"CosineSimilarityStaticBroadcastModule_basic",
"CosineSimilarityStaticModule_basic",
"CumsumInputDtypeInt32Module_basic",
[MHLO] Support aten.cumsum op in mhlo backend (#1825)
2023-01-30 13:38:27 +08:00
"CumsumStaticModule_basic",
"CumsumStaticNegativeDimModule_basic",
[Torch Dialect] fold aten.detach (#2021)
2023-04-18 23:59:14 +08:00
"DetachModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"DivFloatModule_basic",
"DivIntModule_basic",
"DropoutEvalFloatModule_basic",
"DropoutEvalIntModule_basic",
"ElementwiseAbsFloatModule_basic",
"ElementwiseAbsIntModule_basic",
[Stablehlo] Support AtenPowScalarOp, AtenTanOp, AtenAsinhOp, AtenAcoshOp, AtenAtanhOp, Atan2Op (#3233)
2024-04-26 15:47:44 +08:00
"ElementwiseAcoshIntModule_basic",
"ElementwiseAcoshModule_basic",
"ElementwiseAsinhIntModule_basic",
"ElementwiseAsinhModule_basic",
"ElementwiseAtanhIntModule_basic",
"ElementwiseAtanhModule_basic",
"ElementwiseAtan2TensorFloatStaticModule_basic",
"ElementwiseAtan2TensorIntStaticModule_basic",
"ElementwiseAtan2FloatIntStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
"ElementwiseAddScalar_TensorLiteralInt32_Module_basic",
"ElementwiseAtenIsinfOpModule_basic",
"ElementwiseAtenIsneginfOpModule_basic",
"ElementwiseAtenIsposinfOpModule_basic",
Add e2e support for aten logical or/and/xor/not ops (#1761)
2023-01-04 10:11:25 +08:00
"ElementwiseAtenLogicalAndOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseAtenLogicalNotOpModule_basic",
"ElementwiseAtenLogicalNotOpPromoteModule_basic",
"ElementwiseAtenLogicalOrOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseAtenLogicalXorOpPromoteBroadcastStaticShapeModule_basic",
Add more mhlo basic converters (#1628) * Add more mhlo basic converters * remove unused pinnedMemory constraints * refine naming
2022-11-24 14:28:34 +08:00
"ElementwiseAtenWhereSelfModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseBinaryStaticShapeModule_basic",
[LINALG][TOSA][MHLO] Add e2e support for aten bitwise ops (#1753)
2023-01-12 06:40:03 +08:00
"ElementwiseBitwiseAndStaticShapeModule_basic",
"ElementwiseBitwiseNotInt32Module_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseBitwiseNotInt64Module_basic",
[LINALG][TOSA][MHLO] Add e2e support for aten bitwise ops (#1753)
2023-01-12 06:40:03 +08:00
"ElementwiseBitwiseOrStaticShapeModule_basic",
"ElementwiseBitwiseXorStaticShapeModule_basic",
[stablehlo] add aten left/right shift op conversion support (#3234)
2024-04-26 09:20:49 +08:00
"ElementwiseBitwiseLeftShiftInt64Module_basic",
"ElementwiseBitwiseLeftShiftInt8Module_basic",
"ElementwiseBitwiseRightShiftInt64Module_basic",
"ElementwiseBitwiseRightShiftInt8Module_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseCeilModule_basic",
[Torch] emit aten.celu and decompose it (#3247) CELU(x)=max(0,x)+min(0,α∗(exp(x/α)−1))
2024-04-28 17:23:40 +08:00
"ElementwiseCeluStaticModule_basic",
Add AtenClampOp conversion pattern to MHLO (#1356) Add AtenClampOp conversion pattern to MHLO
2022-09-16 15:09:21 +08:00
"ElementwiseClampMaxModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseClampMinModule_basic",
[stablehlo] add aten.clamp.Tensor op conversion support (#3185)
2024-04-19 10:55:27 +08:00
"ElementwiseClampMinTensorFloatModule_basic",
"ElementwiseClampMinTensorIntModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseClampModule_basic",
[stablehlo] add aten.clamp.Tensor op conversion support (#3185)
2024-04-19 10:55:27 +08:00
"ElementwiseClampTensorFloatModule_basic",
"ElementwiseClampTensorIntModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseClampTensorInt8Module_basic",
"ElementwiseCloneChannelsLastMemoryFormatModule_basic",
"ElementwiseCloneContiguousModule_basic",
"ElementwiseCloneModule_basic",
"ElementwiseCosModule_basic",
Added 2 Ops: Floor divide scalar and Floor divide scalar mode (#3156) - Added linalg lowering for `AtenFloorDivideScalarOp` - Needed `AtenDivScalarModeOp` for the decomp. - Added linalg lowering for `AtenDivScalarModeOp` - Moved linalg payload logic to `createDivModePayload()` since the logic was nearly identical for both `AtenDivScalarModeOp` and `AtenDivTensorModeOp`. Just a template function - Added `AtenDivScalarModeOp` lowering for stablehlo Pytorch's [`torch.floor_divide()`](https://pytorch.org/docs/stable/generated/torch.floor_divide.html) in a previous version (for a reason unknown to me) preformed a truncation instead of "floor". The already implemented op `AtenFloorDivideTensorOp` was done before this change. However, this wasn't caught because our testcases only tested positive floor division. I changed this to floor as well as adding a few test cases.
2024-04-16 04:45:10 +08:00
"ElementwiseDivTensorRoundingModeFloorStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncStaticModule_basic",
"ElementwiseDivTensorRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeFloorStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncStaticModule_basic",
"ElementwiseDivScalarRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncIntStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseErfModule_basic",
[MLIR][MHLO] Add AtenFrobeniusNormDimOp and add its conversion pattern to MHLO and linalg (#1306) * Add aten.frobenius_norm.dim op and init its conversion pattern to linalg and MHLO, * run symbolic-shape-optimization before hlo-legalize-to-linalg to fit more mhlo e2e tests.
2022-09-08 10:15:36 +08:00
"ElementwiseExpModule_basic",
[stablehlo] add aten.expm1 op conversion support (#3199)
2024-04-22 10:20:49 +08:00
"ElementwiseExpm1IntModule_basic",
"ElementwiseExpm1Module_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseFloorIntModule_basic",
"ElementwiseFloorModule_basic",
[stablehlo] add aten.fmod.Tensor op conversion support (#3198)
2024-04-21 08:39:36 +08:00
"ElementwiseFmodTensor_Float_basic",
"ElementwiseFmodTensor_Int_Float_basic",
"ElementwiseFmodTensor_Int_basic",
[MLIR][TORCH] Add support for tanh approximation for Gelu op (#2941) Fixes https://github.com/nod-ai/SHARK-Turbine/issues/461 Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-02-27 21:56:01 +08:00
"ElementwiseGeluApproximateTanhModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ElementwiseGeluModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseLeakyReluStaticModule_basic",
[MLIR][MHLO] Add AtenFrobeniusNormDimOp and add its conversion pattern to MHLO and linalg (#1306) * Add aten.frobenius_norm.dim op and init its conversion pattern to linalg and MHLO, * run symbolic-shape-optimization before hlo-legalize-to-linalg to fit more mhlo e2e tests.
2022-09-08 10:15:36 +08:00
"ElementwiseLogModule_basic",
[Stablehlo] add AtenLog10Op, AtenLog2Op lowering to stablehlo (#3208)
2024-04-23 19:06:55 +08:00
"ElementwiseLog10Module_basic",
"ElementwiseLog2Module_basic",
"ElementwiseLog10IntModule_basic",
"ElementwiseLog2IntModule_basic",
[Torch Dialect] fix aten.nan_to_num's decomposition when inf=None (#3530) also add shape infer in decomposition, see https://github.com/llvm/torch-mlir/issues/3312
2024-07-11 08:46:40 +08:00
"ElementwiseNanToNumWithNoneModule_Basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseNanToNumModule_Basic",
"ElementwiseNeFloatTensorStaticModule_basic",
"ElementwiseNeIntTensorStaticModule_basic",
[MLIR][MHLO] Add AtenFrobeniusNormDimOp and add its conversion pattern to MHLO and linalg (#1306) * Add aten.frobenius_norm.dim op and init its conversion pattern to linalg and MHLO, * run symbolic-shape-optimization before hlo-legalize-to-linalg to fit more mhlo e2e tests.
2022-09-08 10:15:36 +08:00
"ElementwiseNegModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseOrTensorStaticShapeModule_basic",
[Torch] Support Aten__And__ScalarOp (#3114)
2024-04-08 20:24:17 +08:00
"ElementwiseAndScalarStaticShapeModule_basic",
[Stablehlo] Support AtenPowScalarOp, AtenTanOp, AtenAsinhOp, AtenAcoshOp, AtenAtanhOp, Atan2Op (#3233)
2024-04-26 15:47:44 +08:00
"ElementwisePowScalarModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwisePowTensorBroadcastStaticModule_basic",
"ElementwisePowTensorStaticModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ElementwisePreluStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseReciprocalModule_basic",
"ElementwiseReluModule_basic",
[stablehlo] add aten.remainder.Tensor op conversion support (#3197)
2024-04-21 00:03:37 +08:00
"ElementwiseRemainderTensorModule_Float_basic",
[TorchToLinalg] Fix torch.aten.remainder for negative operands (#3581) Closes #3575 The PyTorch remainder operator is meant to compute the Python modulus operator entrywise: https://pytorch.org/docs/stable/generated/torch.remainder.html#torch.remainder In python the modulus operator is meant to always return a result with the same sign as the divisor: https://docs.python.org/3/reference/expressions.html#binary-arithmetic-operations In other words, torch.aten.remainder should return a Python-style modulus instead of a C-style modulus. However the remainder operator was simply translated into arith.ModSI or arith.ModF, which both effectively compute the C-style modulus. Now the lowering has been modified so that the modulus operator works properly with negative numbers, both in the dividend, and the divisor.
2024-08-13 23:47:21 +08:00
"ElementwiseRemainderTensorModule_Float_NegativeDividend_basic",
[stablehlo] add aten.remainder.Tensor op conversion support (#3197)
2024-04-21 00:03:37 +08:00
"ElementwiseRemainderTensorModule_Int_Float_basic",
"ElementwiseRemainderTensorModule_Int_basic",
[Torch] Emit rrelu and decompose it (#3250) as title
2024-06-03 19:25:52 +08:00
"ElementwiseRreluEvalStaticModule_basic",
"ElementwiseRreluTrainStaticModule_basic",
[Torch][MHLO] Decompose aten.copy op. Lower aten.rsqrt & sigmoid to mhlo. (#1734)
2022-12-22 10:13:59 +08:00
"ElementwiseRsqrtModule_basic",
"ElementwiseSigmoidModule_basic",
[MHLO] move AtenTanhOp to ConvertAtenUnaryFPOnlyPatten and add sin/cos/ceil/floor pattern (#1847)
2023-02-07 03:14:26 +08:00
"ElementwiseSinModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseSqrtModule_basic",
[Stablehlo] Support AtenPowScalarOp, AtenTanOp, AtenAsinhOp, AtenAcoshOp, AtenAtanhOp, Atan2Op (#3233)
2024-04-26 15:47:44 +08:00
"ElementwiseTanIntModule_basic",
"ElementwiseTanModule_basic",
[Torch] decompose AtenLerpTensorOp (#3251) as title
2024-06-03 15:25:09 +08:00
"ElementwiseTernaryStaticShapeModule_basic",
Replace empty_like && empty_memory_format with full/full_like (#1398) * Replace empty_like && empty_memory_format with full/full_like * fix broadcast rank0 tensor
2022-09-23 10:24:36 +08:00
"ElementwiseToDtypeF32ToI64Module_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ElementwiseToDtypeI64ToI8Module_basic",
"ElementwiseToDtypeIdentityModule_basic",
"ElementwiseUnaryModule_basic",
[Stablehlo] support dynamic shape when convert aten.fill.Scalar (#2349)
2023-07-27 18:35:25 +08:00
"EmptyLikeMemoryFormatModule_basic",
"EmptyLikeModule_defaultDtype",
"EmptyLikeModule_falsePinMemory",
"EmptyLikeModule_float",
"EmptyLikeModule_int",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"EmptyModule_contiguous",
"EmptyModule_defaultDtype",
"EmptyModule_falsePinMemory",
"EmptyModule_float",
"EmptyModule_int",
"EmptyStridedModule_basic",
"EqIntModule_basic",
Replace empty_like && empty_memory_format with full/full_like (#1398) * Replace empty_like && empty_memory_format with full/full_like * fix broadcast rank0 tensor
2022-09-23 10:24:36 +08:00
"ExpandAsIntModule_basic",
Support fake_quantize_per_tensor_affine_cachemask (#3477) Add a new op with shape/dtypes and decompose into `fake_quantize_per_tensor_affine` when the second result is unused. The xfail_set change is on ONNX because torch cannot export this op to ONNX.
2024-06-21 15:15:31 +08:00
"FakeQuantizePerTensorAffineCachemaskModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"FakeQuantizePerTensorAffineModule_basic",
"FakeQuantizePerTensorAffineRoundToEvenModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"Fill_TensorFloat64WithFloat32Static_basic",
[Stablehlo] support dynamic shape when convert aten.fill.Scalar (#2349)
2023-07-27 18:35:25 +08:00
"Fill_TensorFloat64WithFloat32_basic",
"Fill_TensorFloat64WithFloat64_basic",
Add AtenFillScalarOp Stablehlo support (#2108) Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-12 07:41:46 +08:00
"Fill_TensorFloat64WithInt64Static_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"Fill_TensorFloat64WithInt64_basic",
"FlattenRank0Module_basic",
[Stablehlo] add conversion for AtenFlipOp (#2163)
2023-06-15 10:27:34 +08:00
"FlipModuleStaticShape_basic",
"FlipNegativeIndexModule_basic",
Replace empty_like && empty_memory_format with full/full_like (#1398) * Replace empty_like && empty_memory_format with full/full_like * fix broadcast rank0 tensor
2022-09-23 10:24:36 +08:00
"FullLikeModuleDefaultDtype_basic",
"FullLikeModuleFalsePinMemory_basic",
"FullLikeModuleFloat2D_basic",
"FullLikeModuleFloat3DStatic_basic",
"FullLikeModuleFloat3D_basic",
"FullLikeModuleInt2DStatic_basic",
"FullLikeModuleInt2D_basic",
"FullLikeModuleInt3D_basic",
"FullModuleDefaultDtype_basic",
"FullModuleFalsePinMemory_basic",
"FullModuleFloat2D_basic",
"FullModuleFloat3D_basic",
"FullModuleInt2D_basic",
"FullModuleInt3D_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
Added GeluBackward: MHLO support (#1725)
2022-12-21 20:09:43 +08:00
"GeluBackwardModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"GluStaticModule_basic",
[Stablehlo] Add lowering of GridSampler Op (#3084) Inspired by PyTorch decompositions.py. See https://github.com/pytorch/pytorch/blob/ec58f1f74ebcec744d2ab90ad34abd09c1018e92/torch/_decomp/decompositions.py#L3923-L4086 Only support paddingMode=0 or 1 and interpolationMode=0 or 1
2024-06-07 16:06:07 +08:00
"GridSamplerBasic1_basic",
"GridSamplerBasic2_basic",
"GridSamplerBasic3_basic",
"GridSamplerBasic4_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"GtFloatIntModule_basic",
"GtIntModule_basic",
Implement lowering of torch.aten.hstack (#3563)
2024-09-11 19:11:47 +08:00
"HstackBasicComplexModule_basic",
"HstackBasicFloatModule_basic",
"HstackBasicIntFloatModule_basic",
"HstackBasicIntModule_basic",
[Stablehlo] Add `AtenIndexTensor` StableHlo support (#2107) * Add AtenIndexTensor StableHlo support * clean up * Empty commit, trigger test * try to debug hanging test * fix segfulat * fix bad include --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-25 02:13:57 +08:00
"IndexTensorMultiIndexStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"IndexTensorStaticModule_basic",
"IntFloatModule_basic",
"IsFloatingPointFloat_True",
"IsFloatingPointInt_False",
Add E2E support for LeakyRelu and LeakyReluBackward ops (#1733) Co-authored-by: srirammaswamy <srirammaswamy@gmail.com>
2023-01-04 00:30:16 +08:00
"LeakyReluBackwardStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"LenStrModule_basic",
"LiftFreshCopyModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"LinspaceDtypeModule_basic",
"LinspaceEmptyModule_basic",
"LinspaceModule_basic",
"LinspaceOneSizeModule_basic",
"LinspaceTwoSizeModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"MaskedFillScalarFloatValueStaticModule_basic",
"MaskedFillScalarIntValueStaticModule_basic",
[Torch] Add AtenMaskedFillTensorOp support (#3561)
2024-07-26 15:32:13 +08:00
"MaskedFillTensorIntValueStaticModule_basic",
[Torch] Decompose AtenMaskedScatterOp (#3353) Co-authored-by: Yuanqiang Liu <liuyuanqiang.yqliu@bytedance.com>
2024-05-16 15:27:25 +08:00
"MaskedScatterStaticBasic_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"Matmul4dStatic_basic",
"Matmul_2d",
"Matmul_dot",
"Matmul_matvec",
"Matmul_vecmat",
[Stablehlo] Enhance broadcast pattern in matmul Ops (#3161) To pass test "MatmulStaticBroadcast_basic" in stablehlo: ```python class MatmulStaticBroadcast(torch.nn.Module): def __init__(self): super().__init__() @export @annotate_args([ None, ([4, 1, 6, 7], torch.float32, True), ([8, 1, 5, 7, 6], torch.float32, True), ]) def forward(self, lhs, rhs): return torch.matmul(lhs, rhs) @register_test_case(module_factory=lambda: MatmulStaticBroadcast()) def MatmulStaticBroadcast_basic(module, tu: TestUtils): module.forward(tu.rand(4, 1, 6, 7), tu.rand(8, 1, 5, 7, 6)) ```
2024-04-16 10:10:36 +08:00
"MatmulStaticBroadcast_basic",
MaxPool1d lowering to linalg (#3295) Co-authored-by: root <root@i32b01216.sqa.eu95>
2024-05-11 00:35:26 +08:00
"MaxPool1dStaticModule_basic",
"MaxPool1dEmptyStrideStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"MaxPool2dStaticModule_basic",
[Stablehlo] Improve the lowering of pool op in stablehlo (#3259) 1. Handle case stride == None 2. add avgpool3d maxpool1d maxpool3d lowering
2024-05-01 00:06:13 +08:00
"MaxPool2dEmptyStrideStaticModule_basic",
"MaxPool3dStaticModule_basic",
"MaxPool3dEmptyStrideStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"MeanDimAllReduceModule_basic",
fix #1626 return type mismatch (#1634)
2022-11-23 15:02:41 +08:00
"MeanDimEmptyDimModule_basic",
[Stablehlo] Fix AtenSumDimIntListOp when dim==None (#3216) as titile
2024-04-24 11:25:46 +08:00
"MeanDimNoneDimModule_basic",
[MLIR][MHLO] Add AtenFrobeniusNormDimOp and add its conversion pattern to MHLO and linalg (#1306) * Add aten.frobenius_norm.dim op and init its conversion pattern to linalg and MHLO, * run symbolic-shape-optimization before hlo-legalize-to-linalg to fit more mhlo e2e tests.
2022-09-08 10:15:36 +08:00
"MeanDtypeModule_basic",
fix #1626 return type mismatch (#1634)
2022-11-23 15:02:41 +08:00
"MeanDynamicSizesModule_basic",
"MeanModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"Mlp2LayerModuleNoBias_basic",
"MmDagModule_basic",
"MmModule_basic",
"MmModule_chained",
[MLIR][MHLO] Add AtenFrobeniusNormDimOp and add its conversion pattern to MHLO and linalg (#1306) * Add aten.frobenius_norm.dim op and init its conversion pattern to linalg and MHLO, * run symbolic-shape-optimization before hlo-legalize-to-linalg to fit more mhlo e2e tests.
2022-09-08 10:15:36 +08:00
"MmTanhModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"MoveDimIntModule_basic",
"MoveDimIntNegativeIndexModule_basic",
"MulFloatModule_basic",
Added support for integer to complex conversion (#3604)
2024-08-14 20:43:00 +08:00
"MulFloatModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"MulIntModule_basic",
fix mhlo e2e ci crashes (#1620) * fix mhlo e2e ci crashes * add passed tests * calc dynamic positive dim
2022-11-21 21:50:35 +08:00
"Mv_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"NarrowHorizontalTest2_basic",
"NarrowHorizontalTest_basic",
"NarrowTensorHorizontalModule_basic",
"NarrowTensorVerticalModule_basic",
"NarrowVerticalTest2_basic",
"NarrowVerticalTest_basic",
"NativeDropoutEvalFloatModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
[StableHlo] Support AtenEmptyMemoryFormatOp (#2092) * add empty conversion * clean up * add tests --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-19 10:07:35 +08:00
"NewEmptyModuleDefaultDtype_basic",
"NewEmptyModuleFalsePinMemory_basic",
"NewEmptyModuleFloat2D_basic",
"NewEmptyModuleFloat3D_basic",
"NewEmptyModuleInt2D_basic",
"NewEmptyModuleInt3D_basic",
"NewEmptyModuleLayoutIntDtype_basic",
"NewEmptyModuleNonDefaultFloatDtype_basic",
"NewEmptyModuleNonDefaultIntDtype_basic",
"NewEmptyStridedModuleDefaultDtype_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"NewFullModuleDefaultDtype_basic",
"NewFullModuleFalsePinMemory_basic",
"NewFullModuleFloat3DStatic_basic",
"NewFullModuleFloat3D_basic",
"NewFullModuleInt2D_basic",
"NewFullModuleInt3D_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
"NewOnesModuleDefaultDtype_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"NewOnesModuleFalsePinMemory_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
"NewOnesModuleFloat2D_basic",
"NewOnesModuleFloat3D_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"NewOnesModuleInt2D_basic",
"NewOnesModuleInt3D_basic",
"NewZerosModuleDefaultDtype_basic",
"NewZerosModuleFalsePinMemory_basic",
"NewZerosModuleFloat2D_basic",
"NewZerosModuleFloat3D_basic",
"NewZerosModuleInt2D_basic",
"NewZerosModuleInt3D_basic",
[TOSA] Add support for AtenZerosOp 0/strided layout (#1983)
2023-03-30 22:08:20 +08:00
"NewZerosStaticModuleLayoutStrided_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"NumpyTRank0Module_basic",
"NumpyTRank1Module_basic",
"NumpyTRank2Module_basic",
"NumpyTRankNDynamicModule_basic",
"NumpyTRankNStaticModule_basic",
"OnesLikeModule_defaultDtype",
"OnesLikeModule_falsePinMemory",
"OnesLikeModule_float",
"OnesLikeModule_int",
"OnesModuleCPUDevice_basic",
"OnesModuleDefaultDtype_basic",
"OnesModuleFalsePinMemory_basic",
"OnesModuleFloat_basic",
"OnesModuleInt_basic",
[stablehlo] Support ConstantPadNdOp in stablehlo (#3211) as title
2024-04-24 14:15:11 +08:00
"PadModule_basic",
"PadWithNoneValModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"Permute0RankModule_basic",
"PermuteModule_basic",
"PermuteNegativeIndexModule_basic",
[stablehlo] Support PrimsCollapseOp and PrimsSplitDimOp in stablehlo (#3230)
2024-04-29 17:40:30 +08:00
"PixelShuffleModuleStaticRank3Int64_basic",
"PixelShuffleModuleStaticRank4Float32_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"PowIntFloatModule_basic",
[Torch] Fix PrimListUnpackOp::getCanonicalizationPatterns (#3140) Fix the case PrimListUnpackOp's result num is not equal to PrimList length. See the following example: ```python def forward(self, x): if len(x.shape) == 5: b0, t, c0, h0, w0 = x.shape b, c, h, w = torch.mul(b0, t), c0, h0, w0 else: b1, c1, h1, w1 = x.shape b, c, h, w = b1, c1, h1, w1 res = torch.reshape(x, [b, c, h, w]) return res ``` Without this fix, the following error message will occur： ``` /root/torch-mlir/externals/llvm-project/mlir/lib/IR/PatternMatch.cpp:118: virtual void mlir::RewriterBase::replaceOp(mlir::Operation *, mlir::ValueRange): Assertion `op->getNumResults() == newValues.size() && "incorrect # of replacement values"' failed. ```
2024-04-11 19:48:49 +08:00
"PrimListUnpackNumMismatchModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsConvertElementTypeModule_basic",
[Torch] Emit and decompose prims.iota op (#3132)
2024-04-22 10:45:01 +08:00
"PrimsIotaModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
"RandIntDtypeModule_basic",
"RandIntLowDtypeModule_basic",
"RandIntLowModule_basic",
"RandIntModule_basic",
"RandIntPinMemoryModule_basic",
[stablehlo] Support aten.any and aten.all lowering (#3217)
2024-04-25 11:15:52 +08:00
"ReduceAllFloatModule_basic",
"ReduceAllIntModule_basic",
"ReduceAllBoolModule_basic",
"ReduceAnyFloatModule_basic",
"ReduceAnyIntModule_basic",
"ReduceAnyBoolModule_basic",
[stablehlo] Fix test stablehlo e2e test suite (#3093) There is an issue with stablehlo's linalg compilation. Canonicalization appears to cleanup the issues until we can determine what in mlir/stablehlo is the source of the issue.
2024-04-02 12:40:00 +08:00
"ReduceAmaxMultiDim_basic",
"ReduceAmaxOutOfOrderDim_basic",
"ReduceAmaxSingleDim_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ReduceFrobeniusNormModule_basic",
[stablehlo] Fix test stablehlo e2e test suite (#3093) There is an issue with stablehlo's linalg compilation. Canonicalization appears to cleanup the issues until we can determine what in mlir/stablehlo is the source of the issue.
2024-04-02 12:40:00 +08:00
"ReduceMaxAllDims_basic",
"ReduceMaxAlongDimNegative_basic",
"ReduceMaxAlongDimSignedInt_basic",
"ReduceMaxAlongDim_basic",
"ReduceMaxFloatModule_basic",
"ReduceMaxSignedIntModule_basic",
"ReduceMaxUnsignedIntModule_basic",
"ReduceMinFloatModule_basic",
"ReduceMinSignedIntModule_basic",
"ReduceMinUnsignedIntModule_basic",
"ReduceSumDimIntListDtypeFloatModule_basic",
"ReduceSumDimIntListDtypeIntModule_basic",
"ReduceSumDimIntListElementTypeBoolModule_basic",
"ReduceSumDimIntListEmptyDimModule_basic",
"ReduceSumDimIntListFloatModule_basic",
"ReduceSumDimIntListIntModule_basic",
"ReduceSumDtypeFloatModule_basic",
"ReduceSumDtypeIntModule_basic",
"ReduceSumElementTypeBoolModule_basic",
"ReduceSumFloatModule_basic",
"ReduceSumSignedIntModule_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
"ReduceSumUnsignedIntModule_basic",
[Torch] Support AtenProdOp on linalg and stablehlo (#3215)
2024-04-24 11:14:04 +08:00
"ReduceProdFloatModule_basic",
"ReduceProdDtypeFloatModule_basic",
"ReduceProdElementTypeBoolModule_basic",
"ReduceProdUnsignedIntModule_basic",
"ReduceProdSignedIntModule_basic",
"ReduceProdDtypeIntModule_basic",
[Torch] Add decomposition of RepeatInterleaveSelfInt Op (#3075) Decomposition RepeatInterleaveSelfInt with following ops: ```python def my_repeat_interleave(input, repeats, dim=None): if dim is None: # Flatten the input and then repeat return input.flatten().unsqueeze(-1).tile((1, repeats)).flatten() else: # Calculate the shape after repeat expanded_shape = list(input.shape) expanded_shape[dim] *= repeats # Repeat the tensor along the specified dimension repeat_shape = [1] * (input.dim() + 1) repeat_shape[dim + 1] = repeats input = input.unsqueeze(-1) # Tile and then reshape tiled = torch.tile(input, repeat_shape) # Rearrange and reshape repeated = tiled.reshape(*expanded_shape) return repeated ``` I passed the tests of stablehlo and linalg. When testing onnx, strange things happened. In torch-mlir's CI **torch_nightly** and my own environment(torch==2.4.0.dev20240318+cpu), it can **pass the pass**. In torch-mlir's CI **torch_stable**, it **failed**. The test case is `RepeatInterleaveSelfIntNoDimModule_basic`, the result shape should be [120]. ```python class RepeatInterleaveSelfIntNoDimModule(torch.nn.Module): def __init__(self): super().__init__() @export @annotate_args([ None, ([3, 4, 5], torch.float32, True), ]) def forward(self, x): return x.repeat_interleave(2) @register_test_case(module_factory=lambda: RepeatInterleaveSelfIntNoDimModule()) def RepeatInterleaveSelfIntNoDimModule_basic(module, tu: TestUtils): module.forward(tu.rand(3, 4, 5)) ``` The error log is as follows: ``` Unexpected outcome summary: (onnx) ****** Failed tests - 1 tests FAIL - "RepeatInterleaveSelfIntNoDimModule_basic" @ trace item #0 - call to "forward" @ output of call to "forward" ERROR: shape (torch.Size([6, 4, 5])) is not equal to golden shape (torch.Size([120])) ``` @rsuderman Would you please help me check what's wrong with my PR? Thanks a lot.
2024-04-18 06:27:51 +08:00
"RepeatInterleaveSelfIntModule_basic",
"RepeatInterleaveSelfIntNoDimModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ReturnThreeTensorFloat32_basic",
"ReturnTwoTensorF32I64_basic",
"RollModule_basic",
[LINALG] Implement lowering of torch.aten.rot90 (#3551)
2024-09-06 13:06:17 +08:00
"Rot90BasicModule_basic",
"Rot90MultipleRotationsModule_basic",
"Rot90NegativeEvenRotationsModule_basic",
"Rot90NegativeOddRotationsModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"RsubInt0d_NumToTensor_Module_basic",
"ScalarConstantTupleModule_basic",
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
"ScalarTensorDefaultDtypeModule_basic",
"ScalarTensorFloat32Module_basic",
"ScalarTensorInt32Module_basic",
"ScalarTensorInt64Module_basic",
"SliceModule_basic",
"SliceNegIdxModule_basic",
"SliceOutOfLowerBoundStartIndexModule_basic",
"SliceOutOfUpperBoundIndexModule_basic",
"SliceOutOfUpperBoundIndexStaticModule_basic",
"SliceSizeTwoStepModule_basic",
"SliceStartEqEndModule_basic",
"SliceStaticModule_basic",
"SliceWholeTensorModule_basic",
"SortIntListReverse_basic",
"SortIntList_basic",
[stablehlo] Support PrimsCollapseOp and PrimsSplitDimOp in stablehlo (#3230)
2024-04-29 17:40:30 +08:00
"SplitDimStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"SplitTensorGetItem_Module_basic",
"SplitTensorLastSmallerModule_basic",
"SplitTensorListUnpackModule_basic",
"SplitTensorNegativeDimModule_basic",
"SplitWithSizesListUnpackModule_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SqueezeDimModule_identity",
"SqueezeDimModule_unitDim",
"SqueezeModule_allUnitDim",
"SubFloatModule_basic",
"SubIntModule_basic",
"TModuleRank0_basic",
"TModuleRank1_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
"TModuleRank2_basic",
[Torch] Add folder for AtenIntOp, AtenFloatOp (#3189) See unit test below: ``` // CHECK-LABEL: func.func @torch.aten.tensor.float( // CHECK-NEXT: torch.vtensor.literal(dense<1.000000e+01> : tensor<f32>) : !torch.vtensor<[],f32> func.func @torch.aten.tensor.float() -> !torch.vtensor<[],f32> { %none = torch.constant.none %false = torch.constant.bool false %float1.000000e01 = torch.constant.float 1.000000e+01 %67 = torch.aten.tensor.float %float1.000000e01, %none, %none, %false : !torch.float, !torch.none, !torch.none, !torch.bool -> !torch.vtensor<[],f32> return %67 : !torch.vtensor<[],f32> } // CHECK-LABEL: func.func @torch.aten.tensor.int( // CHECK-NEXT: torch.vtensor.literal(dense<45> : tensor<si32>) : !torch.vtensor<[],si32> func.func @torch.aten.tensor.int() -> !torch.vtensor<[],si32> { %none = torch.constant.none %false = torch.constant.bool false %int45 = torch.constant.int 45 %67 = torch.aten.tensor.int %int45, %none, %none, %false : !torch.int, !torch.none, !torch.none, !torch.bool -> !torch.vtensor<[],si32> return %67 : !torch.vtensor<[],si32> } ```
2024-04-19 22:17:06 +08:00
"TensorFloatModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TensorIntModule_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
"TensorLiteralModule_basic",
"TensorOpaqueLiteralModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TensorToBoolZeroRank_basic",
"TensorToFloatZeroRank_basic",
"TensorToIntZeroRank_basic",
"TensorsConcatModule_basic",
[LINALG] Added support for conversion from float to complex. (#3595)
2024-08-07 15:06:48 +08:00
"TensorsConcatComplex128FloatModule_basic",
"TensorsConcatComplex64FloatModule_basic",
Added support for integer to complex conversion (#3604)
2024-08-14 20:43:00 +08:00
"TensorsConcatComplex128IntModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TensorsConcatNegativeDimModule_basic",
"TensorsConcatNegativeDimStaticModule_basic",
"TensorsConcatPromoteDTypeModule_basic",
"TensorsConcatStaticModule_basic",
"TestF16Return_basic",
"TestMultipleTensorAndPrimitiveTypesReturn_basic",
"TestMultipleTensorReturn_basic",
"ToCopyBoolDTypeStaticModule_basic",
fix mhlo e2e ci crashes (#1620) * fix mhlo e2e ci crashes * add passed tests * calc dynamic positive dim
2022-11-21 21:50:35 +08:00
"ToDtypeBoolLayoutNoneStaticModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"ToDtypeLayoutCPUModule_basic",
Replace empty_like && empty_memory_format with full/full_like (#1398) * Replace empty_like && empty_memory_format with full/full_like * fix broadcast rank0 tensor
2022-09-23 10:24:36 +08:00
"ToDtypeLayoutNoneModule_basic",
"ToDtypeLayoutStridedModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"TorchPrimLoopForLikeTensorArgModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TransposeIntModule_basic",
"TransposeIntNegDimsModule_basic",
Implement lowering of torch.aten.triu_indices (#3451) Closes [nod-ai/SHARK-Turbine/issues/709](https://github.com/nod-ai/SHARK-Turbine/issues/709) --------- Co-authored-by: Branko Trifkovic <branko.trifkovic@syrmia.com>
2024-06-22 07:16:38 +08:00
"TriuIndicesModule_basic",
"TriuIndicesAllZerosModule_basic",
"TriuIndicesNegativeOffsetModule_basic",
Implement lowering of torch.aten.tril_indices (#3517)
2024-07-18 21:08:12 +08:00
"TrilIndicesAllZerosModule_basic",
"TrilIndicesModule_basic",
"TrilIndicesNegativeOffsetModule_basic",
"TrilIndicesOfssetGreaterThanRowModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TupleModule_basic",
[torch-dialect] fix aten.type_as op's folder (#2283) [torch-dialect] fix torch.type_as op's folder by decomposing it to prim.dtype + aten.to_dtype
2023-07-20 09:51:58 +08:00
"TypeAsDifferentModule_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"TypeAsSameModule_basic",
Replace empty_like && empty_memory_format with full/full_like (#1398) * Replace empty_like && empty_memory_format with full/full_like * fix broadcast rank0 tensor
2022-09-23 10:24:36 +08:00
"TypeConversionF32ToF64Module_basic",
"TypeConversionF64ToF32Module_basic",
"TypeConversionI1ToF32Module_basic",
"TypeConversionI1ToF64Module_basic",
"TypeConversionI1ToI32Module_basic",
"TypeConversionI1ToI64Module_basic",
"TypeConversionI32ToI64Module_basic",
"TypeConversionI64ToI32Module_basic",
[Stablehlo] add refbackend (#2712)
2024-02-16 01:08:48 +08:00
"UnsafeView1DFoldModule_basic",
"View1DFoldModule_basic",
"ZeroFloat32Module_basic",
"ZeroInt32Module_basic",
"ZeroInt64Module_basic",
"ZerosLikeModule_defaultDtype",
"ZerosLikeModule_falsePinMemory",
"ZerosLikeModule_float",
"ZerosLikeModule_int",
"ZerosModuleDefaultDtype_basic",
"ZerosModuleFalsePinMemory_basic",
"ZerosModuleFloat2D_basic",
"ZerosModuleFloat3D_basic",
"ZerosModuleInt2D_basic",
"ZerosModuleInt3D_basic",
[Stablehlo] add stablehlo-canonicalize-dynamism when lowering (#3097) so that many stablehlo e2e testcases could pass
2024-04-02 22:47:24 +08:00
"AtenEmbeddingBagStaticModule_basic",
"AtenEyeMModuleCPUDevice_basic",
"AtenEyeMModuleDefaultDtype_basic",
"AtenEyeMModuleFalsePinMemory_basic",
"AtenEyeMModuleFloat2D_basic",
"AtenEyeMModuleInt2D_basic",
"AtenEyeModuleCPUDevice_basic",
"AtenEyeModuleDefaultDtype_basic",
"AtenEyeModuleFalsePinMemory_basic",
"AtenEyeModuleFloat2D_basic",
"AtenEyeModuleInt2D_basic",
"AtenInstanceNormModule_basic",
"AtenLinalgCrossBroadcast_basic",
"AtenLinalgCrossCustomDim_basic",
"AtenLinalgCrossFloat_basic",
"AtenLinalgCrossInt_basic",
"AtenLinalgCrossNegativeDim_basic",
"BucketizeTensorStaticFloatModule_basic",
"BucketizeTensorStaticModule_basic",
"DropoutTrainStaticShapeModule_basic",
"ElementwiseWhereScalarOtherStaticModule_basic",
"ElementwiseWhereScalarSelfStaticModule_basic",
"EmbeddingModule1DIndices_basic",
"EmbeddingModuleF16_basic",
"EmbeddingModuleI32Static_basic",
"EmbeddingModuleI32_basic",
"EmbeddingModuleI64_basic",
"GatherStaticModule_basic",
"IndexSelectDynamicIndexSizeModule_basic",
"IndexSelectNegativeDimModule_basic",
"IndexSelectSingleIdxModule_basic",
"IndexSelectTwoIdxModule_basic",
"IndexSelectWholeDimensionModule_basic",
"IndexSelectWholeTensorModule_basic",
"IndexTensorStaticContiguousWithNoneModule_basic",
"IndexTensorStaticNonContiguousWithNoneModule_basic",
"LayerNormLastDimModule_basic",
"LayerNormModule_basic",
"LayerNormNormalizeOverAllDimsModule_basic",
"MaxPool2dWithIndicesStaticModule_basic",
"MeanDimAllReduceKeepdimModule_basic",
"NativeDropoutTrainStaticShapeModule_basic",
"NativeLayerNormModule4D_basic",
"NativeLayerNormModule_basic",
"NormalizeModule_basic",
"PrimsSqueezeModule_basic",
"RandModule_basic",
"ReduceSumDimIntListKeepDimNegativeDimStaticModule_basic",
"SelectIntNegativeDimAndIndexStaticModule_basic",
"SqueezeDimModule_static",
"SqueezeModule_static",
"TriuBroadcastModule_basic",
"TriuModule_basic",
"UnbindIntGetItem_Module_basic",
"UnbindIntListUnpack_Module_basic",
"UniformStaticShapeModule_basic",
[Stablehlo] add stablehlo-aggressive-simplification in e2e test (#3109) * so that more stablehlo e2e testcases would pass.
2024-04-07 10:48:11 +08:00
"ArangeStartOutViewModule_basic",
"ConvolutionBackwardModule2DStrided_basic",
"EinsumStaticContractRhsModule_basic",
"EinsumStaticFourDimensionModule_basic",
"EinsumStaticModule_basic",
"EinsumStaticWithEllipsisSlicingAndBroadcastModule_basic",
"EinsumStaticWithEllipsisSlicingModule_basic",
"FlattenStaticModule_basic",
"GroupNormModule_basic",
"GroupNormNoWeightAndBiasModule_basic",
"NativeGroupNormModule_basic",
"RepeatModule_basic",
"ReshapeAliasCollapseModule_basic",
"ReshapeAliasExpandModule_basic",
"ReshapeAsModule_basic",
"ReshapeExpandModule_basic",
"TileBigDimsSizeModule_basic",
"TileSmallDimsSizeModule_basic",
"UnflattenIntNegativeOneDimStaticModule_basic",
"UnflattenIntNegativeOneSizeStaticModule_basic",
"UnflattenIntStaticModule_basic",
"UnflattenStaticModule_basic",
"UniformNoCorrelationModule_basic",
"UnsafeViewCollapseModule_basic",
"UnsafeViewDynamicExpandModule_basic",
"UnsafeViewExpandModule_basic",
"ViewCollapseInferredDimModule_basic",
"ViewCollapseModule_basic",
"ViewCollapseOnesMiddleModule_basic",
"ViewDynamicExpandCollapseModule_basic",
"ViewDynamicExpandModule_basic",
"ViewExpandCollapseModule_basic",
"ViewExpandCollapseWithOnesModule_basic",
"ViewExpandDynamicDimModule_basic",
"ViewExpandInferredDimModule_basic",
"ViewExpandModule_basic",
"ViewExpandOnesBeforeAndAfterModule_basic",
"ViewExpandOnesMiddleModule_basic",
"ViewExpandOnesModule_basic",
"ViewNegativeStaticModule_basic",
"ViewNoChange1dModule_basic",
"ViewNoChange2dModule_basic",
"ViewNoChange3dModule_basic",
"ViewNoChangeStaticModule_basic",
"ViewOffsetBackwardTestStaticModule_basic",
"ViewOffsetTestStaticModule_basic",
"ViewTwoFiveThreeStaticModule_basic",
"ViewTwoToThreeStaticModule_basic",
[Stablehlo] lowering aten.log1p to stablehlo.log_plus_one (#3110)
2024-04-07 17:01:58 +08:00
"ElementwiseLog1pModule_basic",
[Torch Dialect] canonicalize aten.sign to aten.sgn (#3112) * `aten.sign` is a sub-set of `aten.sgn` (`aten.sgn` support complex type).
2024-04-08 20:05:42 +08:00
"ElementwiseSgnModule_basic",
"ElementwiseSignIntModule_basic",
[Stablehlo] lowering asin, acos, atan (#3207) * lowering asin, acos and atan to chlo ops.
2024-04-23 16:24:53 +08:00
"ElementwiseAcosModule_basic",
"ElementwiseAsinModule_basic",
"ElementwiseAtanTensorFloatModule_basic",
[Stablehlo] support unary ops which promote to floating point (#3209) * promote input to output element-type when lowering to stablehlo, so that it could satisfy stablehlo's type constraints. * split promote-to-fp unary ops from fp-only unary ops.
2024-04-23 17:57:12 +08:00
"ElementwiseAcosIntModule_basic",
"ElementwiseAsinIntModule_basic",
"ElementwiseAtanTensorIntModule_basic",
"ElementwiseCosIntModule_basic",
"ElementwiseExpIntModule_basic",
"ElementwiseLogIntModule_basic",
"ElementwiseRsqrtIntModule_basic",
"ElementwiseSigmoidIntModule_basic",
"ElementwiseSinIntModule_basic",
"ElementwiseSqrtIntModule_basic",
"ElementwiseUnaryIntModule_basic",
[Stablehlo] support lowering sinh & cosh to stablehlo (#3213)
2024-04-23 19:54:58 +08:00
"ElementwiseCoshIntModule_basic",
"ElementwiseCoshModule_basic",
"ElementwiseSinhIntModule_basic",
"ElementwiseSinhModule_basic",
[Torch] support aten.trunc (#3219) decompose `trunc(x)` to `sign(x) * floor(abs(x))`
2024-04-24 14:32:33 +08:00
"ElementwiseTruncIntModule_basic",
"ElementwiseTruncModule_basic",
[Torch] emit aten.log_sigmoid and decompose it to log(sigmoid) (#3246)
2024-04-28 11:47:43 +08:00
"ElementwiseLogSigmoidModule_basic",
[torch] emit aten.Softshrink and aten.Hardshrink (#3248) as title
2024-05-08 15:20:45 +08:00
"ElementwiseHardshrinkStaticModule_basic",
"ElementwiseSoftshrinkStaticModule_basic",
Implement lowering of torch.aten.renorm (#3388) Closes [nod-ai/SHARK-Turbine/issues/689](https://github.com/nod-ai/SHARK-Turbine/issues/689) --------- Co-authored-by: Branko Trifkovic <branko.trifkovic@syrmia.com>
2024-06-18 01:40:57 +08:00
"RenormModuleFloat16_basic",
"RenormModuleFloat32NegativeDim_basic",
"RenormModuleFloat32_basic",
[MHLO] Init end to end unit tests (#1223)
2022-08-23 16:47:21 +08:00
}
Bump llvm to 585523750e2bbe374d1cb3bf4ff9d53de29b9593 (#3613) Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-09 00:36:10 +08:00
STABLEHLO_CRASHING_SET = {
"IndexPutWithNoneAndBroadcastModule_basic",
"ReduceMaxAlongDimUnsignedInt_basic",
"ReduceMinAlongDimUnsignedInt_basic",
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
# LLVM ERROR: Failed to infer result type(s)
"ElementwiseClampMinTensorFloatModule_basic",
"ElementwiseClampMinTensorIntModule_basic",
"ElementwiseClampTensorFloatModule_basic",
"ElementwiseClampTensorIntModule_basic",
}
TOSA_CRASHING_SET = {
# Runtime op verification: Out of bounds access
"IndexTensorNegativeIndexModule_basic",
Bump llvm to 585523750e2bbe374d1cb3bf4ff9d53de29b9593 (#3613) Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-09 00:36:10 +08:00
}
[Stablehlo] Add converter to stablehlo for aten.(Int,Float,Bool).Tensor op (#2340) [Stablehlo] Add converter to stablehlo for aten.(Int,Float,Bool).Tensor op and configure crashing e2e sets for stablehlo backend.
2023-07-29 21:55:49 +08:00
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
FX_IMPORTER_TOSA_CRASHING_SET = {
"IndexTensorNegativeIndexModule_basic",
"InterpolateDynamicModule_scales_recompute_bilinear",
"InterpolateDynamicModule_sizes_bilinear",
"InterpolateDynamicModule_sizes_nearest",
"InterpolateStaticModule_scales_bilinear_align_corners",
"UpSampleNearest2d_basic",
"UpSampleNearest2dStaticSize_basic",
"UpSampleNearest2dDynamicSize_basic",
"UpSampleNearest2dDynamicFactor_basic",
"UpSampleNearest2dStaticFactor_basic",
}
Add a basic TOSA E2E backend. We lower through linalg-on-tensors and use RefBackend to run it. This adds enough support for a "tanh" op. Adding more ops should be fairly mechanical now that things are wired up. Run with: ``` ./tools/torchscript_e2e_test.sh -c tosa ``` The backend structure is very similar to linalg-on-tensors based E2E backends and is a nice parallel (see `tosa_backend.py`). Actually, this forced a nice refactoring to the layering here. We removed `torchscript-module-to-linalg-on-tensors-backend-pipeline` and instead require separately running ``` torchscript-function-to-torch-backend-pipeline,torch-backend-to-linalg-on-tensors-backend-pipeline ``` This highlights the step that lowers to the "torch backend contract" of cleaned up `torch` dialect ops is a critical step in the lowering. Going forward, that is the key load-bearing contract of the torch-mlir project, not the linalg-on-tensors backend contract. Recommended review order: - `TorchToTosa.cpp` / `TorchToTosa/basic.mlir` - `python/torch_mlir_e2e_test/torchscript/configs/tosa_backend.py` and the new `utils.py` file there. - `python/torch_mlir_e2e_test/tosa_backends/linalg_on_tensors.py` and `abc.py` in that directory for the TOSA backend e2e interface. - other misc mechanical changes
2021-10-08 10:07:03 +08:00
# Write the TOSA set as a "passing" set as it is very early in development
# and very few tests work yet.
Bump llvm-project to f1b922188ead5ca492c8d8edd47921b013a22ae0. Includes a fix to use `add_mlir_public_c_api_library` for Torch-MLIR's CAPI library, which is now required (note: upstream sample has it the right way). Disabled a TOSA test per discussion: https://github.com/llvm/torch-mlir/issues/379
2021-10-26 02:43:21 +08:00
TOSA_PASS_SET = {
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
"AtenTrilStaticModule_basic",
"AtenTrilWithNegDiagonalStaticModule_basic",
"AtenTrilWithPosDiagonalStaticModule_basic",
[Torch Dialect] improve argmax/argmin's decomposition to support keep… (#3514) …dim=True when dim=None
2024-07-02 09:08:57 +08:00
"ArgmaxKeepdimModule_basic",
[Torch] Add support for Meshgrid (#3462)
2024-06-14 23:59:08 +08:00
"MeshgridIndexingIJ_basic",
"MeshgridIndexingXY_basic",
"Meshgrid_basic",
[ONNX] Fix AveragePool attributes support (#3235) Issues was found here https://github.com/nod-ai/SHARK-Turbine/issues/643 - [ONNX] Fix padding attributes for onnx.AveragePool - [Linalg] Add countIncludePad false support for AtenAvgPool1/2dOp - [Linalg] Add an avg_pool2d countIncludePad False e2e tests - [Linalg] Fix conflict with AtenAvgPool3dOp - [Linalg] Fix e2e crash with AtenAvgPool1dOp - [Linalg] Add dynamic dim support for AtenAvgPool2dOp - [Linalg] Fix AvgPool2dDivisorOverrideModule crash
2024-06-13 03:16:43 +08:00
"AvgPool2dCountIncludePadFalseStaticModule_basic",
[Torch] support recompose of aten.split.with_sizes and aten.tensor_sp… (#3401) …lit.sections * support recompose to aten.split.with_sizes and aten.tensor_split.sections * fix recompose of aten.chunk
2024-05-31 09:56:47 +08:00
"TensorSplitSections_GetItemModule_basic",
"TensorSplitSections_ListUnpackModule_basic",
Implement lowering of torch.aten.atleast_1d (#3498) This operator is necessary in order to implement torch.aten.vstack. Which will be added in a future PR.
2024-07-17 20:50:30 +08:00
"Atleast1dModule0dInput_basic",
"Atleast1dModule1dInput_basic",
Implement lowering of aten.atleast_2d (#3546) This operator is needed to implement aten.vstack, which will be submitted in a subsequent PR
2024-08-14 21:22:31 +08:00
"Atleast2dModule0dInput_basic",
"Atleast2dModule1dInput_basic",
"Atleast2dModule2dInput_basic",
[Torch] fix aten.linear's decomposition (#3391) * support aten.linear with more rank.
2024-05-27 15:49:50 +08:00
"AtenLinear2D_basic",
"AtenLinear3DBias_basic",
Bump llvm and stablehlo (#3377) * bump llvm to 1e5f29af81a5f6fda308074f6345b9fba4faa71c * bump stablehlo to c44d9af8d4879adccf1054cb61a53377ae5898cb
2024-05-22 23:28:45 +08:00
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
"ElementwiseDivTensorFloatModule_basic",
"ElementwiseMulTensorFloatModule_basic",
"ElementwiseWhereScalarSelfStaticModule_basic",
"GroupNormModule_basic",
"GroupNormNoWeightAndBiasModule_basic",
"NativeGroupNormModule_basic",
[Torch] emit aten.dot and canonicalize it to aten.matmul (#3361) * canonicalize `aten.dot` to `aten.matmul`
2024-05-18 22:45:14 +08:00
"AtenDotModule_basic",
[Linalg] [Stablehlo] Promote type for compare scalar op (#3306)
2024-05-10 02:20:06 +08:00
"ElementwiseFloatTensorGtIntScalarModule_basic",
[Torch] emit aten.log_sigmoid and decompose it to log(sigmoid) (#3246)
2024-04-28 11:47:43 +08:00
"ElementwiseLogSigmoidModule_basic",
[Torch] decompose AtenLerpTensorOp (#3251) as title
2024-06-03 15:25:09 +08:00
"ElementwiseTernaryStaticShapeModule_basic",
[Torch] support aten.trunc (#3219) decompose `trunc(x)` to `sign(x) * floor(abs(x))`
2024-04-24 14:32:33 +08:00
"ElementwiseTruncModule_basic",
"ElementwiseTruncIntModule_basic",
[Torch Dialect] canonicalize aten.sign to aten.sgn (#3112) * `aten.sign` is a sub-set of `aten.sgn` (`aten.sgn` support complex type).
2024-04-08 20:05:42 +08:00
"ElementwiseSgnModule_basic",
"ElementwiseSignIntModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"AdaptiveAvgPool2dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool2dUnitOutputSizeStaticModule_basic",
[Torch] Add general AdaptiveAvgPool2dOp decompose support (#3111) Previously, it could only handle the situations where outputsize == (1, 1) or outputsize == (input_H, input_W). Now it supports all situations where input_H % output_H== 0 && input_W % output_W == 0
2024-04-11 17:02:59 +08:00
"AdaptiveAvgPool2dOutputSizeDivisibleByInputStaticModule_basic",
[Torch] Add support for static uneven divisible AdaptiveAvgPool2d (#3566) The static uneven divisible AdaptiveAvgPool2d means that although the input size is not an integer multiple of ouput size, but the kernel and stride size can also be fixed (not dynamic). The derivation logic of kernel and stride size is consistent with torch/_decomp/decomposations.py:adaptive_avg_pool2d as described in the following: 1. Stride Size Firstly , derive the start index in each reduce operation according to the output size (`n`), `start_index = ([0, 1, ..., n - 1] * input_size) // output_size`. For each index `k`, if `k * (input_size % output_size) < output_size`, then the current and previous stride keeps the same as `input_size // output_size`. So suppose `(n-1) * (input_size % output_size) < output_size`, the stride in the whole AdaptiveAvgPool2d process keeps static, as `input_size // output_size`. 2. Kernel Size torch/_decomp/decomposations.py:adaptive_avg_pool2d calculates a static kernel size when the input/output sizes satisfy either of the two conditions, `input_size % output_size == 0` or `output_size % (input_size % output_size) == 0`. Here if `input_size % output_size == 0`, then the kernel size equals `input_size // output_size`, otherwise `input_size // output_size + 1.`
2024-08-01 11:37:53 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
lowered addcmul and addcdiv to linalg
2021-11-25 06:01:48 +08:00
"AddCDivModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"AddCDiv_Module_basic",
"AddCMulModule_basic",
"AddCMul_Module_basic",
"Add_Module_basic",
"AliasModule_basic",
"ArangeDtypeFloatModule_basic",
"ArangeIntModule_basic",
"ArangeNegativeStartIntModule_basic",
"ArangeStartIntModule_basic",
"ArangeStartNegativeStepIntModule_basic",
"ArangeStartOutModule_basic",
"ArangeStartOutViewModule_basic",
"ArangeStartStepIntModule_basic",
"ArangeZeroElementOutputModule_basic",
Add support for torch arange float module (#2749) Added Support for float dtype in in torch.arange in TOSA Dialect This resolves the following issue :- https://github.com/llvm/torch-mlir/issues/2762 The following test cases are passing after this change 1. ArangeDtypeIntModule_basic 2. ArangeFloatModule_basic 3. ArangeNegativeStartFloatModule_basic 4. ArangeStartFloatModule_basic 5. ArangeStartNegativeStepFloatModule_basic 6. ArangeStartOutDtypeModule_basic 7. ArangeStartStepFloatModule_basic --------- Co-authored-by: James Newling <james.newling@gmail.com>
2024-02-28 05:40:55 +08:00
"ArangeDtypeIntModule_basic",
"ArangeFalsePinMemoryModule_basic",
"ArangeFloatModule_basic",
"ArangeNegativeStartFloatModule_basic",
"ArangeStartFloatModule_basic",
"ArangeStartNegativeStepFloatModule_basic",
"ArangeStartOutDtypeModule_basic",
"ArangeStartStepFloatModule_basic",
Dynamic size support for flatten (#3005) Added support for dynamic shapes in `flattenusingints` op in tosa dialect. Due to this some Argmax tests pass This PR fixes this issue https://github.com/llvm/torch-mlir/issues/3004 The following tests pass after this PR ``` 1. "ArgmaxIntModule_basic" 2. "ArgmaxIntModule_multiple_maxs" 3. "ArgmaxModule_basic" ```
2024-03-20 06:19:29 +08:00
"ArgmaxIntModule_basic",
"ArgmaxIntModule_multiple_maxs",
"ArgmaxModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ArgmaxModule_keepDim",
"AtenComplex64Module_basic",
"AtenEyeMModuleCPUDevice_basic",
"AtenEyeMModuleDefaultDtype_basic",
"AtenEyeMModuleFalsePinMemory_basic",
"AtenEyeMModuleFloat2D_basic",
"AtenEyeModuleCPUDevice_basic",
"AtenEyeModuleDefaultDtype_basic",
"AtenEyeModuleFalsePinMemory_basic",
"AtenEyeModuleFloat2D_basic",
"AtenRoundIntModule_basic",
OnnxToTorch support for onnx.InstanceNormalization op (#2710) https://github.com/nod-ai/SHARK-Turbine/issues/327
2024-02-19 22:23:48 +08:00
"AtenInstanceNormModule_basic",
Add lowering for aten.to.device (#1107)
2022-08-11 07:24:02 +08:00
"AtenToDeviceModule_basic",
[Torch] Support Aten_CastFloatOp. (#3115) By canonicalize Aten_CastFloatOp into AtenToDtypeOp
2024-04-09 11:06:53 +08:00
"Aten_CastFloatModule_basic",
[Torch] Add support for Aten__Or__BoolOp (#3574)
2024-07-31 17:23:53 +08:00
"TrueFalseOrBoolOpModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"BaddbmmBroadcast1DInputModule_basic",
"BaddbmmBroadcast2DInputModule_basic",
"BaddbmmDynamicModule_basic",
"BaddbmmStaticModule_basic",
"BaddbmmWithAlphaBetaModule_basic",
"BaddbmmWithAlphaModule_basic",
"BaddbmmWithBetaModule_basic",
"BatchNorm1DModule_basic",
"BatchNorm1DStaticShapeModule_basic",
"BatchNorm1DWith2DInputModule_basic",
"BatchNorm2DModule_basic",
"BatchNorm3DModule_basic",
[MLIR][TORCH] Add different dtype support for aten.bmm op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-09-11 20:58:59 +08:00
"BmmFloatModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"BoolTensorHandleSignless_basic",
"BoolTensorReturnFalseModule_basic",
"BoolTensorReturnMixedModule_basic",
"BoolTensorReturnTrueModule_basic",
"BroadcastListConstructWithMinusOneModule_basic",
"BroadcastToSameRankStaticModule_basic",
"BroadcastZeroRankInputStaticModule_basic",
"BucketizeTensorStaticFloatModule_basic",
"BucketizeTensorStaticModule_basic",
[Torch Dialect] add fold pattern for aten.clone (#2804)
2024-01-31 09:43:21 +08:00
"CloneModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ChunkListUnpackUneven_Module_basic",
"ChunkListUnpack_Module_basic",
"ConstantBoolParameterModule_basic",
"ConstantPad2dStaticModule_basic",
"ConstantPadNdModule_basic",
"ConstantPadNdPartialStaticModule_basic",
"ConstantPadNdStaticModule_basic",
"ContiguousModule_basic",
"Conv2dBiasNoPaddingModule_basic",
"Conv2dNoPaddingModule_basic",
"Conv2dWithPaddingDilationStrideModule_basic",
"Conv2dWithPaddingDilationStrideStaticModule_basic",
"Conv2dWithPaddingDilationStrideStaticModule_depthwise",
"Conv2dWithPaddingDilationStrideStaticModule_depthwise_multiplier",
"Conv2dWithPaddingModule_basic",
"Convolution2DStaticModule_basic",
torch-to-tosa lowering support for AtenLinalgVectorNormOp (#2734) This PR add torch-to-tosa lowering support for AtenLinalgVectorNormOp e2e test: python -m e2e_testing.main --config=tosa LIT tests: cmake --build build --target tools/torch-mlir/all --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-01-19 04:32:23 +08:00
"CosineSimilarityStaticModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"DetachModule_basic",
"DropoutEvalFloatModule_basic",
"DropoutEvalIntModule_basic",
"DropoutModule_basic",
[Torch Dialect] Support Einsum Op (#2230) As title, support torch.aten.einsum op Right now only support Static Shape, because of the known issue, the fixed solution is here: https://github.com/llvm/torch-mlir/pull/2154 Co-authored-by: Jiawei Wu [wujiawei.aml@bytedance.com](mailto:wujiawei.aml@bytedance.com)
2023-12-10 12:30:37 +08:00
"EinsumStaticContractRhsModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"EinsumStaticFourDimensionModule_basic",
"EinsumStaticModule_basic",
[Torch] support decompose aten.einsum with ellipsis slicing (#3056)
2024-03-28 03:42:10 +08:00
"EinsumStaticWithEllipsisSlicingModule_basic",
"EinsumStaticWithEllipsisSlicingAndBroadcastModule_basic",
Add lowering support for math::AbsIOp (#2875) There is no lowering support for math::AbsIOp, so if the operand is an integer type, it will fail to lower to math::AbsFOp since the op operand #0 must be floating-point-like.
2024-02-09 06:53:40 +08:00
"ElementwiseAbsFloatModule_basic",
"ElementwiseAbsIntModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseAddModule_basic",
"ElementwiseAddScalarFloatModule_basic",
"ElementwiseAddScalarInt64Module_basic",
"ElementwiseAddScalarInt8Module_basic",
"ElementwiseAddScalarIntModule_basic",
"ElementwiseAddScalar_TensorLiteralInt32_Module_basic",
"ElementwiseAtenDivIntScalarModule_basic",
lower torch.aten.isinf to linalg (#2638) Co-authored-by: Rob Suderman <rob.suderman@gmail.com>
2023-12-29 09:20:32 +08:00
"ElementwiseAtenIsinfOpModule_basic",
[Torch Dialect] support aten.isneginf, aten.isposinf, aten.nan_to_num (#2743)
2024-01-16 14:29:34 +08:00
"ElementwiseAtenIsneginfOpModule_basic",
"ElementwiseAtenIsposinfOpModule_basic",
Handle torch.none type in tosa.clamp op (#2739) This PR updates the torch-to-tosa conversion with following changes: - Support torch.none as min/max input argument for tosa.clamp op - Support negative value as start index for tosa.slice op - Add tosa.logical_or lowering support e2e test: python -m e2e_testing.main --config=tosa LIT tests: cmake --build build --target tools/torch-mlir/all --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-01-12 02:36:48 +08:00
"ElementwiseAtenLogicalOrOpBrodcastModule_basic",
"ElementwiseAtenLogicalOrOpDiffArgs1Module_basic",
"ElementwiseAtenLogicalOrOpDiffArgs2Module_basic",
"ElementwiseAtenLogicalOrOpDiffArgs3Module_basic",
"ElementwiseAtenLogicalOrOpModule_basic",
"ElementwiseAtenLogicalOrOpNegativeModule_basic",
"ElementwiseAtenLogicalOrOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseAtenLogicalOrOpRandomFloatModule_basic",
"ElementwiseAtenLogicalOrOpRandomModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseAtenWhereSelfModule_basic",
"ElementwiseBinaryModule_basic",
"ElementwiseBinaryStaticShapeModule_basic",
[LINALG][TOSA][MHLO] Add e2e support for aten bitwise ops (#1753)
2023-01-12 06:40:03 +08:00
"ElementwiseBitwiseAndModule_basic",
"ElementwiseBitwiseAndStaticShapeModule_basic",
"ElementwiseBitwiseNotInt32Module_basic",
"ElementwiseBitwiseNotInt64Module_basic",
"ElementwiseBitwiseOrModule_basic",
"ElementwiseBitwiseOrStaticShapeModule_basic",
"ElementwiseBitwiseXorModule_basic",
"ElementwiseBitwiseXorStaticShapeModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseCeilModule_basic",
[Torch] emit aten.celu and decompose it (#3247) CELU(x)=max(0,x)+min(0,α∗(exp(x/α)−1))
2024-04-28 17:23:40 +08:00
"ElementwiseCeluModule_basic",
"ElementwiseCeluStaticModule_basic",
Handle torch.none type in tosa.clamp op (#2739) This PR updates the torch-to-tosa conversion with following changes: - Support torch.none as min/max input argument for tosa.clamp op - Support negative value as start index for tosa.slice op - Add tosa.logical_or lowering support e2e test: python -m e2e_testing.main --config=tosa LIT tests: cmake --build build --target tools/torch-mlir/all --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-01-12 02:36:48 +08:00
"ElementwiseClampMaxModule_basic",
"ElementwiseClampMinModule_basic",
"ElementwiseClampModule_basic",
Bump LLVM to llvm/llvm-project@e5ed7b6e2fd368b722b6359556cd0125881e7638 (#2964)
2024-02-29 06:13:26 +08:00
"ElementwiseClampTensorInt8Module_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseCloneChannelsLastMemoryFormatModule_basic",
"ElementwiseCloneContiguousModule_basic",
"ElementwiseCloneModule_basic",
"ElementwiseDivScalarModule_basic",
[MLIR][TORCH] Extend support for OnnxToLinalg lowering for Dropout and Div op (#2938) Fixes https://github.com/nod-ai/SHARK-Turbine/issues/451, https://github.com/nod-ai/SHARK-Turbine/issues/452
2024-02-27 13:32:05 +08:00
"ElementwiseDivTensorIntegerModule_basic",
"ElementwiseDivTensorUnsignedIntegerModule_basic",
Added 2 Ops: Floor divide scalar and Floor divide scalar mode (#3156) - Added linalg lowering for `AtenFloorDivideScalarOp` - Needed `AtenDivScalarModeOp` for the decomp. - Added linalg lowering for `AtenDivScalarModeOp` - Moved linalg payload logic to `createDivModePayload()` since the logic was nearly identical for both `AtenDivScalarModeOp` and `AtenDivTensorModeOp`. Just a template function - Added `AtenDivScalarModeOp` lowering for stablehlo Pytorch's [`torch.floor_divide()`](https://pytorch.org/docs/stable/generated/torch.floor_divide.html) in a previous version (for a reason unknown to me) preformed a truncation instead of "floor". The already implemented op `AtenFloorDivideTensorOp` was done before this change. However, this wasn't caught because our testcases only tested positive floor division. I changed this to floor as well as adding a few test cases.
2024-04-16 04:45:10 +08:00
"ElementwiseDivScalarIntegerModule_basic",
"ElementwiseDivScalarUnsignedIntegerModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseEluModule_basic",
"ElementwiseEluNonDefaultModule_basic",
"ElementwiseEqBoolScalarModule_basic",
"ElementwiseEqDiffWidthScalarModule_basic",
"ElementwiseEqFloatScalarModule_basic",
"ElementwiseEqFloatTensorModule_basic",
"ElementwiseEqIntScalarModule_basic",
"ElementwiseEqIntTensorModule_basic",
"ElementwiseExpModule_basic",
"ElementwiseFlattenBroadcastModule_basic",
"ElementwiseFloorIntModule_basic",
"ElementwiseFloorModule_basic",
[torch dialect] emit aten.fmax/fmin and add decomposition patterns (#3510)
2024-06-29 00:07:55 +08:00
"ElementwiseFmaxModule_basic",
"ElementwiseFminModule_basic",
Decompose aten.fmod into aten.mul,sub,div etc. (#3689) As titled, create a new decomposition for `aten.fmod.Tensor` to `aten.div`, `aten.trunc`, `aten.mul` and `aten.sub`. Note that we only use `aten.trunc` for floating point operations. This further gets decomposed to `aten.where` etc. by other existing decompositions. This decomposition now makes TOSA pass for a simple model with `aten.fmod` while it makes `stablehlo` fail. For now, we disallow this decomposition for `stablehlo` --------- Co-authored-by: Srinath Avadhanula <srinath.avadhanula@getcruise.com>
2024-09-10 00:00:11 +08:00
"ElementwiseFmodTensor_Float_basic",
"ElementwiseFmodTensor_Int_Float_basic",
"ElementwiseFmodTensor_Int_basic",
Support aten.sign (#2205)
2023-06-11 02:45:35 +08:00
"ElementwiseGeFloatIntScalarModule_basic",
"ElementwiseGeFloatScalarModule_basic",
"ElementwiseGeIntScalarModule_basic",
"ElementwiseGeMixedIntScalarModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseGeluModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseGtFloatScalarModule_basic",
"ElementwiseGtFloatTensorModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseGtIntScalarModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseGtIntTensorModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseGtMixed2ScalarModule_basic",
"ElementwiseIsinfModule_basic",
[Torch Dialect] support aten.isneginf, aten.isposinf, aten.nan_to_num (#2743)
2024-01-16 14:29:34 +08:00
"ElementwiseAtenIsneginfOpModule_basic",
"ElementwiseAtenIsposinfOpModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseIsnanModule_basic",
"ElementwiseLeFloatTensorModule_basic",
"ElementwiseLeIntTensorModule_basic",
"ElementwiseLeakyReluModule_basic",
"ElementwiseLeakyReluModule_basic",
"ElementwiseLeakyReluStaticModule_basic",
Implement lowering of torch.aten.lerp.Scalar (#2773) Closes nod-ai/SHARK-Turbine#356
2024-02-01 01:39:38 +08:00
"ElementwiseLerpScalarIntModule_basic",
"ElementwiseLerpScalarFloatModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseLog2Module_basic",
"ElementwiseLogModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseLtDiffWidthScalarModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseLtFloatScalarModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseLtFloatTensorModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseLtIntScalarModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseLtIntTensorModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseMaxOtherIntModule_basic",
"ElementwiseMaxOtherModule_basic",
"ElementwiseMaximumIntModule_basic",
"ElementwiseMaximumModule_basic",
"ElementwiseMinOtherIntModule_basic",
"ElementwiseMinOtherModule_basic",
"ElementwiseMinimumIntModule_basic",
"ElementwiseMinimumModule_basic",
"ElementwiseMulScalarModule_basic",
"ElementwiseMulScalarModule_float",
"ElementwiseMulScalarModule_float",
"ElementwiseMulScalarModule_int",
"ElementwiseMulTensorIntModule_basic",
[Torch Dialect] add support for AtenIsnanOp (#2170) * add support for mhlo * Add Test for torch.ne * fix torch.ne shape/add static test case * add support for static torch.ne --------- Co-authored-by: root <root@n31-177-039.byted.org>
2023-06-07 10:06:27 +08:00
"ElementwiseNeFloatScalarModule_basic",
"ElementwiseNeFloatTensorModule_basic",
"ElementwiseNeFloatTensorStaticModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseNeIntScalarModule_basic",
[Torch Dialect] add support for AtenIsnanOp (#2170) * add support for mhlo * Add Test for torch.ne * fix torch.ne shape/add static test case * add support for static torch.ne --------- Co-authored-by: root <root@n31-177-039.byted.org>
2023-06-07 10:06:27 +08:00
"ElementwiseNeIntTensorModule_basic",
"ElementwiseNeIntTensorStaticModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseNegModule_basic",
"ElementwiseOrTensorModule_basic",
"ElementwiseOrTensorStaticShapeModule_basic",
"ElementwisePowModule_basic",
[Torch] Support prelu decomposition (#3069)
2024-03-29 08:05:00 +08:00
"ElementwisePreluModule_basic",
"ElementwisePreluStaticModule_basic",
Implement lowering for torch.aten.rad2deg (#3586)
2024-08-14 19:07:28 +08:00
"ElementwiseRad2DegModule_basic",
"ElementwiseRad2DegIntModule_basic",
* [tosa] Support for AtenRsubScalarOp for scalar constants (#531) * [tosa] Support for AtenCeilOp and AtenReciprocalOp * [tosa] Support for comparator ops, Aten[Gt|Lt|Eq][Tensor|Scalar]Op with scalar constant * [tosa] Support for Scalar variants of Aten[Mul|Div|Add|Sub] Ops with scalar constants Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-01-21 02:58:30 +08:00
"ElementwiseReciprocalModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseRelu6Module_basic",
"ElementwiseReluModule_basic",
[TorchToLinalg] Fix torch.aten.remainder for negative operands (#3581) Closes #3575 The PyTorch remainder operator is meant to compute the Python modulus operator entrywise: https://pytorch.org/docs/stable/generated/torch.remainder.html#torch.remainder In python the modulus operator is meant to always return a result with the same sign as the divisor: https://docs.python.org/3/reference/expressions.html#binary-arithmetic-operations In other words, torch.aten.remainder should return a Python-style modulus instead of a C-style modulus. However the remainder operator was simply translated into arith.ModSI or arith.ModF, which both effectively compute the C-style modulus. Now the lowering has been modified so that the modulus operator works properly with negative numbers, both in the dividend, and the divisor.
2024-08-13 23:47:21 +08:00
"ElementwiseRemainderScalarModule_Float_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Float_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_Float_NegativeDivisor_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseRemainderScalarModule_Float_basic",
"ElementwiseRemainderScalarModule_Int_Float_basic",
"ElementwiseRemainderScalarModule_Int_basic",
"ElementwiseRemainderScalarModule_Int_basic",
[Torch] Emit rrelu and decompose it (#3250) as title
2024-06-03 19:25:52 +08:00
"ElementwiseRreluEvalModule_basic",
"ElementwiseRreluEvalStaticModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseRsqrtModule_basic",
[torch] Add support for aten.selu (#2640) Add `aten.selu` operation to `torch` dialect.
2023-12-14 12:28:08 +08:00
"ElementwiseSeluModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseSigmoidModule_basic",
"ElementwiseSignModule_basic",
[torch] emit aten.Softshrink and aten.Hardshrink (#3248) as title
2024-05-08 15:20:45 +08:00
"ElementwiseHardshrinkModule_basic",
"ElementwiseHardshrinkStaticModule_basic",
"ElementwiseSoftshrinkModule_basic",
"ElementwiseSoftshrinkStaticModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseSqrtIntModule_basic",
"ElementwiseSqrtModule_basic",
"ElementwiseSubScalarFloatModule_basic",
[tosa] Support for some ops and fix for Issue #532 (#575) * [tosa] Support for AtenNe[Tensor|Scalar]Op, AtenLog2Op, AtenBitwiseAndTensorOp, AtenSquareOp and AtenThresholdOp * Fix for Issue #532 - Mixed input types for few ops and updated few tests to use i32 instead of i64 Signed-off-by: Anup Gangwar <anup.gangwar@arm.com> Co-authored-by: Anup Gangwar <anup.gangwar@arm.com>
2022-02-12 04:30:02 +08:00
"ElementwiseSubScalarIntModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ElementwiseSubTensorInt8Module_basic",
"ElementwiseToDtypeIdentityModule_basic",
"ElementwiseUnaryModule_basic",
[MLIR][TORCH] Fix aten.unsqueeze op (#1578) The range of the unsqueeze dim is: [-input.dim() - 1, input.dim() + 1), the bug forgets to add 1.
2022-11-15 01:09:15 +08:00
"ElementwiseUnsqueezeBroadcastModule_basic",
[TOSA] Add torch.prim.NumToTensor.Scalar float support (#1802)
2023-04-19 04:36:57 +08:00
"ElementwiseWhereScalarModule_basic",
[Torch Dialect] fix aten.nan_to_num's decomposition when inf=None (#3530) also add shape infer in decomposition, see https://github.com/llvm/torch-mlir/issues/3312
2024-07-11 08:46:40 +08:00
"ElementwiseNanToNumWithNoneModule_Basic",
[Torch Dialect] support aten.isneginf, aten.isposinf, aten.nan_to_num (#2743)
2024-01-16 14:29:34 +08:00
"ElementwiseNanToNumModule_Basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"EmbeddingModule1DIndices_basic",
"EmbeddingModuleI32Static_basic",
"FlattenRank0Module_basic",
"FlattenStaticModule_basic",
Dynamic size support for flatten (#3005) Added support for dynamic shapes in `flattenusingints` op in tosa dialect. Due to this some Argmax tests pass This PR fixes this issue https://github.com/llvm/torch-mlir/issues/3004 The following tests pass after this PR ``` 1. "ArgmaxIntModule_basic" 2. "ArgmaxIntModule_multiple_maxs" 3. "ArgmaxModule_basic" ```
2024-03-20 06:19:29 +08:00
"FlattenDynamicModuleCollapseAll_basic",
[TOSA] Add torch.prim.NumToTensor.Scalar float support (#1802)
2023-04-19 04:36:57 +08:00
"FullLikeModuleFloat3DStatic_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"FullLikeModuleInt2DStatic_basic",
[TOSA] Add torch.prim.NumToTensor.Scalar float support (#1802)
2023-04-19 04:36:57 +08:00
"FullModuleDefaultDtype_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"FullModuleFloat2D_basic",
[TOSA] Add torch.prim.NumToTensor.Scalar float support (#1802)
2023-04-19 04:36:57 +08:00
"FullModuleFloat3D_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"FullModuleInt3D_basic",
"GatherStaticModule_basic",
"GeluBackwardModule_basic",
"GluStaticModule_basic",
"HardTanhIntModule_basic",
"HardTanhModule_basic",
"HardsigmoidModule_basic",
"HardsigmoidRandomModule_basic",
"HardswishModule_basic",
"HardswishRandomModule_basic",
"HardtanhBackward_basic",
"IndexTensorMultiIndexStaticModule_basic",
"IndexTensorStaticModule_basic",
"IscloseStaticModuleTrue_basic",
"IscloseStaticModule_basic",
"LayerNormNormalizeOverAllDimsModule_basic",
"LeakyReluBackwardModule_basic",
"LeakyReluBackwardStaticModule_basic",
"LiftFreshCopyModule_basic",
torch-to-tosa lowering support for AtenLinalgVectorNormOp (#2734) This PR add torch-to-tosa lowering support for AtenLinalgVectorNormOp e2e test: python -m e2e_testing.main --config=tosa LIT tests: cmake --build build --target tools/torch-mlir/all --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-01-19 04:32:23 +08:00
"LinalgVectorNormKeepDimModule_basic",
"LinalgVectorNormModule_basic",
e2e support aten.linalg_norm to aten.linalg_vector_norm (#2953) Add e2d support for `aten.linalg_norm` by decompose it to `aten.linalg_vector_norm`. Lowering to `aten.linalg_matrix_norm` is still unsupported. To Test: `python -m e2e_testing.main -v` --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-03-06 08:31:01 +08:00
"LinalgNormKeepDimModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"MaskedFillScalarDefaultModule_basic",
"MaskedFillScalarIntValueModule_basic",
"MaskedFillScalarIntValueStaticModule_basic",
"MaskedFillTensorIntValueStaticModule_basic",
"Matmul4dStatic_basic",
"Matmul_3d",
"Matmul_dot",
[tosa] Fix TOSA batch matmul lowering to correct transpose ordering (#2959) The corrective transpose at the end is computed incorrectly. Is it actually computin the inverse transpose. Inverting the permutations fixes the issue.
2024-02-29 01:46:58 +08:00
"MatmulStaticBroadcast_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"MaxPool2dEmptyStrideStaticModule_basic",
"MaxPool2dStaticCeilModeTrueModule_basic",
"MaxPool2dStaticModule_basic",
"MeanModule_basic",
"MmDagModule_basic",
"MoveDimIntModule_basic",
"MoveDimIntModule_basic",
"MoveDimIntNegativeIndexModule_basic",
"MseLossNoReductionModule_basic",
"NativeLayerNormModule4D_basic",
[MLIR][TORCH] add E2E support for aten.new_full (#2425) * implement aten.new_full * remove extraneous tests
2023-09-12 22:29:08 +08:00
"NewFullModuleDefaultDtype_basic",
"NewFullModuleFalsePinMemory_basic",
"NewFullModuleFloat2D_basic",
"NewFullModuleFloat3DStatic_basic",
"NewFullModuleFloat3D_basic",
"NewFullModuleInt2DStatic_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"NewOnesModuleDefaultDtype_basic",
"NewOnesModuleFalsePinMemory_basic",
"NewOnesModuleFloat2D_basic",
"NewOnesModuleFloat3D_basic",
"NewOnesModuleInt2D_basic",
"NewOnesModuleInt3D_basic",
"NewZerosModuleDefaultDtype_basic",
"NewZerosModuleFalsePinMemory_basic",
"NewZerosModuleFloat2D_basic",
"NewZerosModuleFloat3D_basic",
"NewZerosModuleInt2D_basic",
"NewZerosModuleInt3D_basic",
"NewZerosStaticModuleLayoutStrided_basic",
torch-to-tosa lowering support for AtenLinalgVectorNormOp (#2734) This PR add torch-to-tosa lowering support for AtenLinalgVectorNormOp e2e test: python -m e2e_testing.main --config=tosa LIT tests: cmake --build build --target tools/torch-mlir/all --------- Co-authored-by: Ze Zhang <ze.zhang@getcruise.com>
2024-01-19 04:32:23 +08:00
"NormalizeModule_basic",
"NormScalarOptDimKeepDimModule_basic",
"NormScalarOptDimModule_basic",
[TOSA] Add torch.prim.NumToTensor.Scalar float support (#1802)
2023-04-19 04:36:57 +08:00
"NumToTensorFloatModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"NumToTensorIntModule_basic",
"NumpyTRank0Module_basic",
"NumpyTRank1Module_basic",
"NumpyTRank2Module_basic",
"NumpyTRankNDynamicModule_basic",
"NumpyTRankNStaticModule_basic",
"OnesModuleCPUDevice_basic",
"OnesModuleDefaultDtype_basic",
"OnesModuleFalsePinMemory_basic",
"OnesModuleFloat_basic",
"OnesModuleInt_basic",
"PadModule_basic",
"PadWithNoneValModule_basic",
"Permute0RankModule_basic",
"PermuteModule_basic",
"PermuteNegativeIndexModule_basic",
[Torch] Fix PrimListUnpackOp::getCanonicalizationPatterns (#3140) Fix the case PrimListUnpackOp's result num is not equal to PrimList length. See the following example: ```python def forward(self, x): if len(x.shape) == 5: b0, t, c0, h0, w0 = x.shape b, c, h, w = torch.mul(b0, t), c0, h0, w0 else: b1, c1, h1, w1 = x.shape b, c, h, w = b1, c1, h1, w1 res = torch.reshape(x, [b, c, h, w]) return res ``` Without this fix, the following error message will occur： ``` /root/torch-mlir/externals/llvm-project/mlir/lib/IR/PatternMatch.cpp:118: virtual void mlir::RewriterBase::replaceOp(mlir::Operation *, mlir::ValueRange): Assertion `op->getNumResults() == newValues.size() && "incorrect # of replacement values"' failed. ```
2024-04-11 19:48:49 +08:00
"PrimListUnpackNumMismatchModule_basic",
[Torch] Emit and decompose prims.iota op (#3132)
2024-04-22 10:45:01 +08:00
"PrimsIotaModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
[stablehlo] Support aten.any and aten.all lowering (#3217)
2024-04-25 11:15:52 +08:00
"ReduceAllBoolModule_basic",
"ReduceAnyBoolModule_basic",
build: update llvm tag to 4546397e (#1502) This commit makes the following changes needed to update bump LLVM: - Replace `linalg.init_tensor` with `tensor.empty` (see: https://reviews.llvm.org/D135129) - Replace `NoSideEffect` with `Pure` (see https://reviews.llvm.org/D135505) - Replace `body` region accessor for `ReduceOp` and `ReduceWindowOp` with `getBody` - Fix incorrect use of `tosa::ReduceSumOp` in `AtenNativeLayerNormOp` conversion pattern. The result type of `tosa::ReduceSumOp` must have the same rank as the input type. (see: https://www.mlplatform.org/tosa/tosa_spec.html#_reduce_sum) Co-authored-by: Ashay Rane <ashay@users.noreply.github.com> Co-authored-by: Ashay Rane <ashay@users.noreply.github.com>
2022-10-18 12:22:53 +08:00
"ReduceSumDimIntListFloatModule_basic",
"ReduceSumDimIntListIntModule_basic",
"ReduceSumDimIntListKeepDimFloatModule_basic",
"ReduceSumDimIntListKeepDimIntModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ReduceSumDimIntListKeepDimNegativeDimStaticModule_basic",
build: update llvm tag to 4546397e (#1502) This commit makes the following changes needed to update bump LLVM: - Replace `linalg.init_tensor` with `tensor.empty` (see: https://reviews.llvm.org/D135129) - Replace `NoSideEffect` with `Pure` (see https://reviews.llvm.org/D135505) - Replace `body` region accessor for `ReduceOp` and `ReduceWindowOp` with `getBody` - Fix incorrect use of `tosa::ReduceSumOp` in `AtenNativeLayerNormOp` conversion pattern. The result type of `tosa::ReduceSumOp` must have the same rank as the input type. (see: https://www.mlplatform.org/tosa/tosa_spec.html#_reduce_sum) Co-authored-by: Ashay Rane <ashay@users.noreply.github.com> Co-authored-by: Ashay Rane <ashay@users.noreply.github.com>
2022-10-18 12:22:53 +08:00
"ReduceSumFloatModule_basic",
"ReduceSumSignedIntModule_basic",
"ReduceSumUnsignedIntModule_basic",
Constant pad nd to tosa (#1933) * implemented lowering torch.aten.constant_pad_nd to tosa * add constant_pad_nd e2e tests to TOSA_PASS_SET * add PadModule_basic & PadWithNoneValModule_basic to TOSA_PASS_SET --------- Co-authored-by: Lisa Liu <lingl@xilinx.com>
2023-03-15 23:42:15 +08:00
"RepeatModule_basic",
[Torch] Add decomposition of RepeatInterleaveSelfInt Op (#3075) Decomposition RepeatInterleaveSelfInt with following ops: ```python def my_repeat_interleave(input, repeats, dim=None): if dim is None: # Flatten the input and then repeat return input.flatten().unsqueeze(-1).tile((1, repeats)).flatten() else: # Calculate the shape after repeat expanded_shape = list(input.shape) expanded_shape[dim] *= repeats # Repeat the tensor along the specified dimension repeat_shape = [1] * (input.dim() + 1) repeat_shape[dim + 1] = repeats input = input.unsqueeze(-1) # Tile and then reshape tiled = torch.tile(input, repeat_shape) # Rearrange and reshape repeated = tiled.reshape(*expanded_shape) return repeated ``` I passed the tests of stablehlo and linalg. When testing onnx, strange things happened. In torch-mlir's CI **torch_nightly** and my own environment(torch==2.4.0.dev20240318+cpu), it can **pass the pass**. In torch-mlir's CI **torch_stable**, it **failed**. The test case is `RepeatInterleaveSelfIntNoDimModule_basic`, the result shape should be [120]. ```python class RepeatInterleaveSelfIntNoDimModule(torch.nn.Module): def __init__(self): super().__init__() @export @annotate_args([ None, ([3, 4, 5], torch.float32, True), ]) def forward(self, x): return x.repeat_interleave(2) @register_test_case(module_factory=lambda: RepeatInterleaveSelfIntNoDimModule()) def RepeatInterleaveSelfIntNoDimModule_basic(module, tu: TestUtils): module.forward(tu.rand(3, 4, 5)) ``` The error log is as follows: ``` Unexpected outcome summary: (onnx) ****** Failed tests - 1 tests FAIL - "RepeatInterleaveSelfIntNoDimModule_basic" @ trace item #0 - call to "forward" @ output of call to "forward" ERROR: shape (torch.Size([6, 4, 5])) is not equal to golden shape (torch.Size([120])) ``` @rsuderman Would you please help me check what's wrong with my PR? Thanks a lot.
2024-04-18 06:27:51 +08:00
"RepeatInterleaveSelfIntNoDimModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ResNet18StaticModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ReshapeAliasCollapseModule_basic",
"ReshapeAliasExpandModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ReshapeAsModule_basic",
"ReshapeCollapseModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ReshapeExpandModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ReturnThreeTensorFloat32_basic",
"ReturnTwoTensorF32I64_basic",
"RsubFloatModule_basic",
"RsubFloatModule_noalpha_basic",
"RsubInt0d_NumToTensor_Module_basic",
Add canonicalize pattern for aten.mul.int and aten.floordiv.int (#3680) This PR add `floordiv` to the `PY_BUILTIN_TO_TORCH_OP`. For `aten.mul.int` and `aten.floordiv.int` ops, we add new Canonicalization Patterns as follow: ``` %1 = torch.aten.mul.int %input, %const-5 %2 = torch.aten.mul.int %1, %const-6 ``` Will be replaced by `torch.aten.mul.int %input, %const-30` And ``` %1 = torch.aten.mul.int %input, %const-5 %2 = torch.aten.floordiv.int %1, %const-5 ``` Will directly return `%input` This PR also relaxes the `float` type constraint in TorchToTosa for the `AtenRsubScalarOp` conversion. To test: `cmake --build build --target check-torch-mlir-all`
2024-09-04 00:13:59 +08:00
"RsubIntModule_basic",
[Torch Dialect] Add support for AtenScalarTensorOp (#2085) * add scalar_tensor op * add dynamo pass test; needs PR2062 * try to fix * Empty commit, trigger test * Empty commit, trigger test * address comments * use dtype function * fix decompose rule * remove unused include * Empty commit, trigger test * fix test * disable ltc * fix dtype --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-06-01 11:38:50 +08:00
"ScalarTensorDefaultDtypeModule_basic",
"ScalarTensorFloat32Module_basic",
"ScalarTensorInt32Module_basic",
"ScalarTensorInt64Module_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"SelectIntNegativeDimAndIndexStaticModule_basic",
"SiluModule_basic",
"SliceOutOfUpperBoundIndexStaticModule_basic",
"SliceStaticModule_basic",
[Torch Dialect] Add split.tensor support + recompose rules (#2102) * add split.tensor support + recompose rules * add e2e test * address comments * address comments * erase op in recomposeOp --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-24 03:43:33 +08:00
"SplitTensorGetItem_Module_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"SplitTensorLastSmallerModule_basic",
[Torch Dialect] Support recompose aten.split.Tensor + prim.ListUnpack (#2192)
2023-06-07 01:38:04 +08:00
"SplitTensorListUnpackModule_basic",
Fixes for split tensor and slice (#2314) * RecomposeComplexOps: Remove dead slice op * lib/Dialect/Torch/IR/TorchOps.cpp: Fold slice ops even when they are on non-value tensors * lib/Conversion/TorchToTosa/TorchToTosa.cpp: Fix slice start/end out of range/none * lib/Dialect/Torch/IR/TorchOps.cpp: AtenSliceTensorOp::fold: Fold slices that go from 0:int_max * More tests for aten.split.Tensor
2023-07-20 15:53:54 +08:00
"SplitTensorNegativeDimModule_basic",
[Torch Dialect] support aten.split_with_sizes (#2431) * [Torch Dialect] support aten.split_with_sizes * update
2023-09-04 09:59:26 +08:00
"SplitWithSizesListUnpackModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"SquareModule_basic",
"SqueezeDimModule_identity",
"SqueezeDimModule_static",
"SqueezeDimModule_unitDim",
"SqueezeModule_allUnitDim",
"SqueezeModule_broadcast",
"SqueezeModule_noUnitDim",
"SqueezeModule_static",
"TModuleRank0_basic",
"TModuleRank1_basic",
"TModuleRank2_basic",
"TanhBackward_basic",
[Torch] Add folder for AtenIntOp, AtenFloatOp (#3189) See unit test below: ``` // CHECK-LABEL: func.func @torch.aten.tensor.float( // CHECK-NEXT: torch.vtensor.literal(dense<1.000000e+01> : tensor<f32>) : !torch.vtensor<[],f32> func.func @torch.aten.tensor.float() -> !torch.vtensor<[],f32> { %none = torch.constant.none %false = torch.constant.bool false %float1.000000e01 = torch.constant.float 1.000000e+01 %67 = torch.aten.tensor.float %float1.000000e01, %none, %none, %false : !torch.float, !torch.none, !torch.none, !torch.bool -> !torch.vtensor<[],f32> return %67 : !torch.vtensor<[],f32> } // CHECK-LABEL: func.func @torch.aten.tensor.int( // CHECK-NEXT: torch.vtensor.literal(dense<45> : tensor<si32>) : !torch.vtensor<[],si32> func.func @torch.aten.tensor.int() -> !torch.vtensor<[],si32> { %none = torch.constant.none %false = torch.constant.bool false %int45 = torch.constant.int 45 %67 = torch.aten.tensor.int %int45, %none, %none, %false : !torch.int, !torch.none, !torch.none, !torch.bool -> !torch.vtensor<[],si32> return %67 : !torch.vtensor<[],si32> } ```
2024-04-19 22:17:06 +08:00
"TensorFloatModule_basic",
"TensorIntModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"TensorLiteralModule_basic",
"TensorOpaqueLiteralModule_basic",
"TensorsConcatNegativeDimStaticModule_basic",
"TensorsConcatStaticModule_basic",
"TestF16Return_basic",
"TestMultipleTensorReturn_basic",
"Threshold1dFloatModule_basic",
"Threshold1dIntI32Module_basic",
"Threshold2dFloatModule_basic",
"Threshold3dFloatModule_basic",
"TileBigDimsSizeModule_basic",
"TileSmallDimsSizeModule_basic",
"ToCopyBoolDTypeStaticModule_basic",
"ToDtypeBoolLayoutNoneStaticModule_basic",
"TransposeIntModule_basic",
"TransposeIntNegDimsModule_basic",
Fix canonicalization of `torch.prim.TupleUnpack`.
2023-07-18 22:32:26 +08:00
"TupleModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"TypeAsSameModule_basic",
"TypePromotionAlphaWiderModule_basic",
"TypePromotionDifferentCategoryModule_basic",
"TypePromotionSameCategoryDifferentWidthModule_basic",
"TypePromotionSameCategoryZeroRankWider_basic",
"TypePromotionZeroRankHigherCategoryModule_basic",
"UnbindIntGetItem_Module_basic",
"UnbindIntListUnpack_Module_basic",
"UnflattenIntNegativeOneDimStaticModule_basic",
"UnflattenIntNegativeOneSizeStaticModule_basic",
"UnflattenIntStaticModule_basic",
"UnflattenStaticModule_basic",
"UnsafeView1DFoldModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"UnsafeViewCollapseModule_basic",
"UnsafeViewDynamicExpandModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"UnsafeViewExpandModule_basic",
"View1DFoldModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ViewCollapseModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ViewCollapseInferredDimModule_basic",
"ViewCollapseOnesMiddleModule_basic",
"ViewDoubleMergeStaticModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ViewDynamicExpandCollapseModule_basic",
"ViewDynamicExpandModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ViewExpandCollapseModule_basic",
"ViewExpandCollapseWithOnesModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ViewExpandDynamicDimModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ViewExpandInferredDimModule_basic",
"ViewExpandModule_basic",
"ViewExpandOnesBeforeAndAfterModule_basic",
"ViewExpandOnesMiddleModule_basic",
"ViewExpandOnesMiddleOppModule_basic",
"ViewExpandOnesModule_basic",
"ViewFiveTestStaticModule_basic",
"ViewNegativeStaticModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ViewNoChange1dModule_basic",
"ViewNoChange2dModule_basic",
"ViewNoChange3dModule_basic",
Sort the TOSA passing test list (#2630) For easier tracking of issues, sort the TOSA passing list. It is still significantly smaller then the XFAIL list would be. Resolves #2620, at least until the xfail list gets smaller than the passing list. Signed-off-by: Eric Kunze <eric.kunze@arm.com>
2023-12-13 06:22:25 +08:00
"ViewNoChangeStaticModule_basic",
"ViewOffsetBackwardTestStaticModule_basic",
"ViewOffsetTestStaticModule_basic",
"ViewTwoFiveThreeStaticModule_basic",
"ViewTwoToThreeStaticModule_basic",
"ZerosModuleDefaultDtype_basic",
"ZerosModuleFalsePinMemory_basic",
"ZerosModuleFloat2D_basic",
"ZerosModuleFloat3D_basic",
"ZerosModuleInt2D_basic",
"ZerosModuleInt3D_basic",
"_LogSoftmaxModuleStable_basic",
[Torch Dialect] support decomposition of aten.linspace (#3006)
2024-03-14 08:28:33 +08:00
"LinspaceModule_basic",
"LinspaceOneSizeModule_basic",
"LinspaceTwoSizeModule_basic",
Fix SCF Forloop fails to convert to linalg when a tensor argument is supplied to the loop block (#3040) Co-authored-by: Rob Suderman <rob.suderman@gmail.com> Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
2024-03-21 02:04:02 +08:00
"TorchPrimLoopForLikeTensorArgModule_basic",
Implement lowering of torch.aten.renorm (#3388) Closes [nod-ai/SHARK-Turbine/issues/689](https://github.com/nod-ai/SHARK-Turbine/issues/689) --------- Co-authored-by: Branko Trifkovic <branko.trifkovic@syrmia.com>
2024-06-18 01:40:57 +08:00
"RenormModuleFloat32NegativeDim_basic",
"RenormModuleFloat32_basic",
[TOSA] Fix Tensor.hacked_twin to support diff size indexes (#3547) - Broadcasts index list tensors - Adds torch.nn.Unfold test Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-07-31 05:32:05 +08:00
"IndexTensorStaticContiguousWithNoneModule_basic",
"IndexTensorStaticNonContiguousWithNoneModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
}
Add make_fx_tosa variant to end2end tests (#2240) * Add make_fx_tosa variant to end2end tests * e2e_testing/xfail_sets.py: Add make_fx_tosa xfail for stable
2023-07-13 21:07:54 +08:00
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
MAKE_FX_TOSA_CRASHING_SET = TOSA_CRASHING_SET | {
# Runtime op verification: static result dims in reassoc group do not divide src dim evenly
"FlattenDynamicModule_basic",
"ReshapeDynamicModule_basic",
"ViewFlattenAndExpandModule_basic",
"ViewSizeDimLedAndFollowedByExpandedOnesModule_basic",
"ViewSizeDimLedByExpandedOnesModule_basic",
}
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
MAKE_FX_TOSA_PASS_SET = (
TOSA_PASS_SET
| {
### Tests additionally passing in make_fx_tosa
[ONNX] Fix AveragePool attributes support (#3235) Issues was found here https://github.com/nod-ai/SHARK-Turbine/issues/643 - [ONNX] Fix padding attributes for onnx.AveragePool - [Linalg] Add countIncludePad false support for AtenAvgPool1/2dOp - [Linalg] Add an avg_pool2d countIncludePad False e2e tests - [Linalg] Fix conflict with AtenAvgPool3dOp - [Linalg] Fix e2e crash with AtenAvgPool1dOp - [Linalg] Add dynamic dim support for AtenAvgPool2dOp - [Linalg] Fix AvgPool2dDivisorOverrideModule crash
2024-06-13 03:16:43 +08:00
"AvgPool2dCountIncludePadFalseStaticModule_basic",
[Torch] fix aten.linear's decomposition (#3391) * support aten.linear with more rank.
2024-05-27 15:49:50 +08:00
"AtenLinear1D_basic",
"AtenLinearMatVec_basic",
"AtenLinearVecMatBias_basic",
Implement lowering of torch.aten.atleast_1d (#3498) This operator is necessary in order to implement torch.aten.vstack. Which will be added in a future PR.
2024-07-17 20:50:30 +08:00
"Atleast1dModule0dInput_basic",
"Atleast1dModule1dInput_basic",
Implement lowering of aten.atleast_2d (#3546) This operator is needed to implement aten.vstack, which will be submitted in a subsequent PR
2024-08-14 21:22:31 +08:00
"Atleast2dModule0dInput_basic",
"Atleast2dModule1dInput_basic",
"Atleast2dModule2dInput_basic",
MaxPool1d lowering to linalg (#3295) Co-authored-by: root <root@i32b01216.sqa.eu95>
2024-05-11 00:35:26 +08:00
"MaxPool1dEmptyStrideStaticModule_basic",
"MaxPool1dStaticCeilModeTrueModule_basic",
"MaxPool1dStaticModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"AdaptiveAvgPool1dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool1dUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool1dStaticEvenMultiple_basic",
"CosineSimilarityModule_basic",
"NativeGroupNormBackwardModule_basic",
"ReduceFrobeniusNormKeepDimModule_basic",
"ReduceFrobeniusNormModule_basic",
"SliceWholeTensorModule_basic",
"TensorFloatModule_basic",
"TensorIntModule_basic",
"AdaptiveAvgPool1dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool1dUnitOutputSizeStaticModule_basic",
"RepeatInterleaveSelfIntModule_basic",
"TorchPrimLoopForLikeTensorArgModule_basic",
"ViewSizeDimFollowedByCollapsedOnesModule_basic",
"ViewSizeDimFollowedByExpandedOnesModule_basic",
"ViewSizeDimLedAndFollowedByCollapsedOnesModule_basic",
"ViewSizeDimLedByCollapsedOnesModule_basic",
"ViewSizeFromOtherTensor_basic",
Implement lowering of torch.aten.renorm (#3388) Closes [nod-ai/SHARK-Turbine/issues/689](https://github.com/nod-ai/SHARK-Turbine/issues/689) --------- Co-authored-by: Branko Trifkovic <branko.trifkovic@syrmia.com>
2024-06-18 01:40:57 +08:00
"RenormModuleFloat32NegativeDim_basic",
"RenormModuleFloat32_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
}
) - {
### Test failing in make_fx_tosa but not in tosa
Add make_fx_tosa variant to end2end tests (#2240) * Add make_fx_tosa variant to end2end tests * e2e_testing/xfail_sets.py: Add make_fx_tosa xfail for stable
2023-07-13 21:07:54 +08:00
# Dynamic shape, has extra unsupported broadcast ops
"Matmul_3d",
[tosa] Fix TOSA batch matmul lowering to correct transpose ordering (#2959) The corrective transpose at the end is computed incorrectly. Is it actually computin the inverse transpose. Inverting the permutations fixes the issue.
2024-02-29 01:46:58 +08:00
"MatmulStaticBroadcast_basic",
Add make_fx_tosa variant to end2end tests (#2240) * Add make_fx_tosa variant to end2end tests * e2e_testing/xfail_sets.py: Add make_fx_tosa xfail for stable
2023-07-13 21:07:54 +08:00
# Unimplemented operator 'aten._index_put_impl_.hacked_twin'
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
build: update llvm tag to 41895843 Summary of changes: - Update tags llvm: 41895843b5915bb78e9d02aa711fa10f7174db43 mhlo: 4726d31f7025da66de0dea709bd56c462edb83c2 Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-07-26 15:30:13 +08:00
# RuntimeError: The size of tensor a (7) must match the size of tensor b (3) at non-singleton dimension 1
"Add_Module_basic",
add e2e support for torch.eye operations (aten.eye, aten.eye.m) (#2478)
2023-11-02 02:23:28 +08:00
# failed to legalize operation 'torch.aten.to.dtype' that was explicitly marked illegal
"AtenEyeModuleInt2D_basic",
"AtenEyeMModuleInt2D_basic",
Advance llvm-project and stablehlo. (#2619) llvm-project: bbd2b08b95fe76bea138c1b03c1cd42ed3ee04df stablehlo: ab709fe48de88c67717abfbd7ef17425eb95ddaf These commits were chosen in order to account for an MLIR API break from https://github.com/llvm/llvm-project/commit/3dbac2c007c114a720300d2a4d79abe9ca1351e7 which required a patch to stablehlo. We integrate a bit beyond that commit to deal with some revert/reapply cycles in the intervening range which were discovered in another downstream. Further, it requires adaptation to the stablehlo API breaks introduced from https://github.com/openxla/stablehlo/pull/1872 which are along for the ride. Since some stablehlo builders were changed to directly take int64_t array refs, also traced that up some call stacks to eliminate some signed/unsigned mismatches that result. Also adds a few TOSA tests to the passing set that seem to work now.
2023-12-08 15:13:42 +08:00
"Conv2dBiasNoPaddingModule_basic",
"Conv2dNoPaddingModule_basic",
"Conv2dWithPaddingDilationStrideModule_basic",
"Conv2dWithPaddingModule_basic",
[Torch] Support prelu decomposition (#3069)
2024-03-29 08:05:00 +08:00
# failed to legalize operation 'torch.operator'
"ElementwisePreluModule_basic",
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"ElementwisePreluStaticModule_basic",
[Torch] emit aten.log_sigmoid and decompose it to log(sigmoid) (#3246)
2024-04-28 11:47:43 +08:00
"ElementwiseLogSigmoidModule_basic",
[Torch] Emit rrelu and decompose it (#3250) as title
2024-06-03 19:25:52 +08:00
# failed to legalize operation 'torch.aten.rrelu_with_noise'
"ElementwiseRreluEvalModule_basic",
"ElementwiseRreluEvalStaticModule_basic",
Implement lowering of torch.aten.hstack (#3563)
2024-09-11 19:11:47 +08:00
# incompatible return type failure for tosa.concat.
"HstackBasicComplexModule_basic",
"HstackBasicFloatModule_basic",
"HstackBasicIntFloatModule_basic",
"HstackBasicIntModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
# Shape Related failures
[Torch] Fix PrimListUnpackOp::getCanonicalizationPatterns (#3140) Fix the case PrimListUnpackOp's result num is not equal to PrimList length. See the following example: ```python def forward(self, x): if len(x.shape) == 5: b0, t, c0, h0, w0 = x.shape b, c, h, w = torch.mul(b0, t), c0, h0, w0 else: b1, c1, h1, w1 = x.shape b, c, h, w = b1, c1, h1, w1 res = torch.reshape(x, [b, c, h, w]) return res ``` Without this fix, the following error message will occur： ``` /root/torch-mlir/externals/llvm-project/mlir/lib/IR/PatternMatch.cpp:118: virtual void mlir::RewriterBase::replaceOp(mlir::Operation *, mlir::ValueRange): Assertion `op->getNumResults() == newValues.size() && "incorrect # of replacement values"' failed. ```
2024-04-11 19:48:49 +08:00
"PrimListUnpackNumMismatchModule_basic",
Bump LLVM to llvm/llvm-project@0030fc4ac74a9ce645adb9d59e108da4d4d11818 (#3079) Co-authored-by: Peiming Liu <peiming@google.com>
2024-04-02 07:34:59 +08:00
"ReshapeExpandModule_basic",
"UnsafeViewCollapseModule_basic",
"UnsafeViewDynamicExpandModule_basic",
"ViewCollapseModule_basic",
"ViewDynamicExpandCollapseModule_basic",
"ViewDynamicExpandModule_basic",
"ViewExpandDynamicDimModule_basic",
"ViewNoChange1dModule_basic",
"ViewNoChange2dModule_basic",
"ViewNoChange3dModule_basic",
[Torch] Add support for static uneven divisible AdaptiveAvgPool2d (#3566) The static uneven divisible AdaptiveAvgPool2d means that although the input size is not an integer multiple of ouput size, but the kernel and stride size can also be fixed (not dynamic). The derivation logic of kernel and stride size is consistent with torch/_decomp/decomposations.py:adaptive_avg_pool2d as described in the following: 1. Stride Size Firstly , derive the start index in each reduce operation according to the output size (`n`), `start_index = ([0, 1, ..., n - 1] * input_size) // output_size`. For each index `k`, if `k * (input_size % output_size) < output_size`, then the current and previous stride keeps the same as `input_size // output_size`. So suppose `(n-1) * (input_size % output_size) < output_size`, the stride in the whole AdaptiveAvgPool2d process keeps static, as `input_size // output_size`. 2. Kernel Size torch/_decomp/decomposations.py:adaptive_avg_pool2d calculates a static kernel size when the input/output sizes satisfy either of the two conditions, `input_size % output_size == 0` or `output_size % (input_size % output_size) == 0`. Here if `input_size % output_size == 0`, then the kernel size equals `input_size // output_size`, otherwise `input_size // output_size + 1.`
2024-08-01 11:37:53 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
Advance llvm-project and stablehlo. (#2619) llvm-project: bbd2b08b95fe76bea138c1b03c1cd42ed3ee04df stablehlo: ab709fe48de88c67717abfbd7ef17425eb95ddaf These commits were chosen in order to account for an MLIR API break from https://github.com/llvm/llvm-project/commit/3dbac2c007c114a720300d2a4d79abe9ca1351e7 which required a patch to stablehlo. We integrate a bit beyond that commit to deal with some revert/reapply cycles in the intervening range which were discovered in another downstream. Further, it requires adaptation to the stablehlo API breaks introduced from https://github.com/openxla/stablehlo/pull/1872 which are along for the ride. Since some stablehlo builders were changed to directly take int64_t array refs, also traced that up some call stacks to eliminate some signed/unsigned mismatches that result. Also adds a few TOSA tests to the passing set that seem to work now.
2023-12-08 15:13:42 +08:00
}
Add make_fx_tosa variant to end2end tests (#2240) * Add make_fx_tosa variant to end2end tests * e2e_testing/xfail_sets.py: Add make_fx_tosa xfail for stable
2023-07-13 21:07:54 +08:00
build: manually update PyTorch version (#3666) Set PyTorch and TorchVision version to nightly release 2024-08-25. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-28 00:28:30 +08:00
if torch_version_for_comparison() < version.parse("2.5.0.dev"):
MAKE_FX_TOSA_PASS_SET = MAKE_FX_TOSA_PASS_SET | {
"ScaledDotProductAttentionDifferentModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
"ScaledDotProductAttentionMaskModule_basic",
"ScaledDotProductAttentionSameModule_basic",
build: manually update PyTorch version (#3666) Set PyTorch and TorchVision version to nightly release 2024-08-25. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-08-28 00:28:30 +08:00
}
[LTC] Tensor[]? support operands type support using partial codegen (#2410) * Tensor[]? support operands type support using partial codegen * aten.index.Tensor support via partial codegen * Add torch.index_put tracing support * Added optional tensor list type support for LTC/TorchMLIR lowering * Added comments Co-authored-by: Gleb Kazantaev <gleb.kazantaev@cerebras.net>
2023-08-30 18:29:39 +08:00
LTC_CRASHING_SET = {
# TODO: update test to move all inputs to the lazy device. Otherwise test fails with:
# Check failed: lazy_tensor Input tensor is not a lazy tensor: CPUBoolType.
"HBC_basic",
}
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
LTC_XFAIL_SET = {
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
"TorchPrimLoopForLikeTensorArgModule_basic" "CollapseAllDimensionsModule_basic",
Support for prims collapse op (lowering to linalg) (#2572) Steps taken: 1) add generator code to torch_ods_gen.py, run update_torch_ods.sh 2) add (custom) shape and type inference generator code to abstract_interp_lib_gen.py, run update_abstract_interp_lib.sh 3) Implement lowering to tensor.collapse_dims. Requires the `start` and `end` values to be constant, else lowering fails 4) Update xfail_sets.py (append to LTC_XFAIL_SET) after running /tools/e2e_test.sh --filter Collapse --verbose -c XX for all support backends (XX). Motivation: - Supporting the collapse operation will be useful for lowering of pixel_shuffle (see Issue #2559)
2023-11-16 00:34:38 +08:00
"CollapseRank1DynamicModule_basic",
"CollapseStaticModule_basic",
"CollapsePartialDynamicModule_basic",
"CollapseFullDynamicModule_basic",
Lowering to linalg of prims split_dim op (#2576) Adds support for lowering to prims split_op. Similar design to collapse op lowering in https://github.com/llvm/torch-mlir/pull/2572, with some small differences, because the split_dim op (in pytorch) is view-changing whereas the collapse is not. The difference means that 1) it must be registered in the function Torch::isViewLikeOp 2) it must be be added to the "expected fail" set for the torch dynamo backend.
2023-11-21 23:56:09 +08:00
"SplitDimStaticModule_basic",
"SplitDimDynamicModule_basic",
Decomposition of aten.pixel_shuffle with static input shape (#2550) For static tests (that is when the shape is know) for example: ``` @annotate_args([None, ([3, 18, 2, 2], torch.float32, True)]) ``` The e2e passes. But only if the replacement op's return type is set as undefined (optional shape and type must be explicitly made unset), otherwise there's a error about the function return type. For dynamic cases, for example if the above is replaced with ``` @annotate_args([None, ([-1, -1, -1, -1], torch.float32, True)]) ``` There is a failure to lower to linalg from torch ("view op explicitly labelled as illegal"). This seems to be because the support for lowering from torch to linalg with dynamic shapes is limited.
2023-11-08 21:52:44 +08:00
"PixelShuffleModuleStaticRank3Int64_basic",
"PixelShuffleModuleStaticRank4Float32_basic",
Improve decomposition of pixel_shuffle (support dynamic shapes) (#2590) The aten.reshape ops in the decomposition are replaced with prims.collapse and prims.split_dim ops, which means that the cases where the lowering of reshape from torch to linalg which are not supported, are avoided. Essentially, by using the collapse and split_dim ops instead of the reshape ops, we are not "losing" the information that the reshapes do not arbitrarily mix dimensions. Which makes lowering easy. 3 additional tests added: - fully dynamic, - dynamic only the spatial dimensions, - dynamic only in the non-spatial dimensions.
2023-11-23 04:31:06 +08:00
"PixelShuffleModuleFullDynamic_basic",
"PixelShuffleModuleSpatiallyDynamic_basic",
"PixelShuffleModuleSpatiallyStatic_basic",
implement aten.conv1d, aten.conv3d, and aten.conv_tbc (#2757) convolution with [time,batch,channel] ordering, as opposed to the default [batch, channel, time]. Currently implementing by transposing the input and output, but may need to get its own implementation in the future because this is supposed to be an op that gives a speedup. This is used by fairseq (https://github.com/facebookresearch/fairseq/issues/172). (in case you were wondering like me, this is different from transposed convolution. Transposed convolution has fractional strides). --------- Co-authored-by: Xida Ren <xida.ren.dev@gmail.com> Co-authored-by: Frederik Harwath <frederik.harwath@amd.com>
2024-01-24 13:30:03 +08:00
"ConvTbcModule_basic",
Blacklist _convolution op (#1048) * Blacklist _convolution op in LTC * Removed duplicate Torch_AtenSelectScatterOp instance from autogen .td * Removed duplicate Torch_AtenSliceScatterOp instance from autogen .td
2022-07-14 01:28:05 +08:00
"_Convolution2DAllFalseModule_basic",
"_Convolution2DBenchmarkModule_basic",
"_Convolution2DCudnnModule_basic",
"_Convolution2DDeterministicModule_basic",
"_Convolution2DTF32Module_basic",
[MLIR][TORCH] Fix CI failure due to failing tests Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-08-22 13:44:44 +08:00
"_ConvolutionDeprecated2DAllFalseModule_basic",
"_ConvolutionDeprecated2DBenchmarkModule_basic",
Fix as_strided/slice symint (#1401) * Fix as_strided symint * Re-enable LTC tests * Re-enable LTC * Add hardtanh shape inference function * Fix slice symint
2022-09-27 00:16:49 +08:00
"_ConvolutionDeprecated2DCudnnModule_basic",
[MLIR][TORCH] Fix CI failure due to failing tests Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-08-22 13:44:44 +08:00
"_ConvolutionDeprecated2DDeterministicModule_basic",
Add aten.pool_max3d support to torch-to-linalg (#2735) Added verification logic to the abstract_interpreter_lib_gen.py Also made some unit tests Initially, I thought we can use `linalg::pooling_ndhwc_max` to help implement this problem. However, on a 5-dimensional matrix it does the pooling on dimensions (2, 3, 4) which is not what we want. We want pooling on dimensions (3, 4, 5). To achieve this, we would need to lower our code using the `linalg` dialect. Turns out the pooling code in `linalg` looks like this. ``` func @max_pooling_ncdhw(%I: memref<?x?x?x?x?xf32>, %K: memref<3xindex>, %O: memref<?x?x?x?x?xf32>, %strides: memref<3xindex>, %dilations: memref<3xindex>) { %c0 = arith.constant 0 : index %c1 = arith.constant 1 : index %N = memref.dim %I, %c0 : memref<?x?x?x?x?xf32> %C = memref.dim %I, %c1 : memref<?x?x?x?x?xf32> %D = memref.dim %I, 2 : memref<?x?x?x?x?xf32> %H = memref.dim %I, 3 : memref<?x?x?x?x?xf32> %W = memref.dim %I, 4 : memref<?x?x?x?x?xf32> %kernel_d = memref.load %K[%c0] : memref<3xindex> %kernel_h = memref.load %K[%c1] : memref<3xindex> %kernel_w = memref.load %K[2] : memref<3xindex> %stride_d = memref.load %strides[%c0] : memref<3xindex> %stride_h = memref.load %strides[%c1] : memref<3xindex> %stride_w = memref.load %strides[2] : memref<3xindex> %dilation_d = memref.load %dilations[%c0] : memref<3xindex> %dilation_h = memref.load %dilations[%c1] : memref<3xindex> %dilation_w = memref.load %dilations[2] : memref<3xindex> linalg.generic { indexing_maps = [ affine_map<(n, c, d, h, w, kd, kh, kw) -> (n, c, d * %stride_d + kd * %dilation_d, h * %stride_h + kh * %dilation_h, w * %stride_w + kw * %dilation_w)>, // Map for input tensor affine_map<(n, c, d, h, w, kd, kh, kw) -> (kd, kh, kw)>, // Map for kernel tensor affine_map<(n, c, d, h, w, kd, kh, kw) -> (n, c, d, h, w)> // Map for output tensor ], iterator_types = ["parallel", "parallel", "parallel", "parallel", "parallel", "reduction", "reduction", "reduction"], doc = "3D Max Pooling NCDHW with Strides, Dilations, and Kernel Size" } ins(%I, %K : memref<?x?x?x?x?xf32>, memref<3xindex>) outs(%O : memref<?x?x?x?x?xf32>) { ^bb0(%input_elem: f32, %kernel_elem: index, %output_elem: f32): %max_val = arith.maxf %input_elem, %output_elem : f32 linalg.yield %max_val : f32 } return } ``` This was implemented based on it's source code with the adjustments mentioned above: https://github.com/llvm/llvm-project/blob/4ca1b5e094280ef1af40412e3cfcb62dc3cf15bc/mlir/include/mlir/Dialect/Linalg/IR/LinalgNamedStructuredOps.yaml#L5647 Issues related to this can be found here https://github.com/nod-ai/SHARK-Turbine/issues/324
2024-01-19 23:39:46 +08:00
"MaxPool3dEmptyStrideStaticModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[Torch Dialect] support AtenArangeStartOutOp in ReduceOpVariants like… (#2563) … AtenBernoulli_FloatOp It fixing case like: `%2110 = torch.aten.arange.start_out %int1, %int1517, %int1, %2109 : !torch.int, !torch.int, !torch.int, !torch.tensor -> !torch.tensor`. `aten.arange.start_out` doesn't have value semantics also, means`%2110` is an alias for %2109. So I decompose it to `aten.arange.start` + `torch.contents.overwrite`. The complex decomposition logic is target to handle cases like view and dtype cast which I add in e2e tests.
2023-11-17 00:51:55 +08:00
"ArangeStartOutViewModule_basic",
Fix empty tensor when select -1 (#1787)
2023-01-18 02:14:14 +08:00
"AtenIntBoolOpModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"BernoulliTensorModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"CeilFloatModule_basic",
"DivFloatModule_basic",
"EqIntModule_basic",
Implement lowering of torch.aten.exponential (#2680) https://github.com/llvm/torch-mlir/issues/2646 Decompose aten.exponential() into: -exp(1-x)/lambda
2023-12-28 12:33:18 +08:00
"ExponentialModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"GeFloatIntModule_basic",
"GeFloatModule_basic",
Resolve remaining LTC CI failures (#1110) * Replace CHECK_EQ with TORCH_CHECK_EQ * Check value of TORCH_MLIR_USE_INSTALLED_PYTORCH during LTC build * Update LTC XFAIL with NewZerosModule ops * Explicitly blacklist _like ops * Automatically blacklist new_/_like ops * Prune away unused Python dependencies from LTC * Add flag to disable LTC * Autogen dummy _REFERENCE_LAZY_BACKEND library when LTC is disabled * Implement compute_shape_var * Removed Var tests from XFAIL Set * XFAIL tests using _local_scalar_dense or index.Tensor * Add StdDim tests to XFAIL set * Autogen aten::cat
2022-07-30 17:54:40 +08:00
"GeIntModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"GtFloatIntModule_basic",
"GtIntModule_basic",
"IndexPutImpl1DFloatAccumulateModule_basic",
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
"IndexPutImpl2DFloatAccumulateModule_basic",
"IndexPutImpl2DFloatNonAccumulateModule_basic",
[MLIR][TORCH] Extend implementation of aten._index_put_impl op. This commits adds the support for cases for index_put_op: 1.) where index is a 2-d tensor. 2.) where indices is a list of tensors and none, with exactly 2 non none tensors along the consecutive dimensions. This commit also adds a utility to compute the broadcast shape given the two input tensors. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-12-08 13:46:54 +08:00
"IndexPutImpl2DIndexModule_basic",
[TOSA] Add aten._index_put_impl support (#2031) Add e2e support by add "tosa-to-scf"
2023-07-18 00:51:24 +08:00
"IndexPutImpl2DNoneIndexStaticModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"IndexPutImpl3DFloatAccumulateModule_basic",
"IndexPutImpl3DFloatNonAccumulateModule_basic",
[MLIR][TORCH] Extend implementation of aten._index_put_impl op. This commits adds the support for cases for index_put_op: 1.) where index is a 2-d tensor. 2.) where indices is a list of tensors and none, with exactly 2 non none tensors along the consecutive dimensions. This commit also adds a utility to compute the broadcast shape given the two input tensors. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-12-08 13:46:54 +08:00
"IndexPutImplIndexWithNoneModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"Matmul_dot",
"MulIntModule_basic",
[MLIR][TORCH] Add E2E support for `aten.[div.int|bitwise_or.Tensor]` ops This commit adds lowering of `aten.div.int` and `aten.bitwise_or.Tensor` ops. Both these ops are required in order to support bloom_560m model. Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-10-06 21:11:52 +08:00
"DivIntModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"NeFloatIntModule_basic",
"NeIntModule_basic",
"QuantizedMLP_basic",
[ONNX] Preliminary Work Towards Supporting QuantizedMLP_basic onnx e2e test (#3089) See the related issues here: [SHARK-Turbine#556](https://github.com/nod-ai/SHARK-Turbine/issues/556) 1. Adds uint8 casting to onnx.Cast op 2. Fixes an issue with onnx.DequantizeLinear when the scale comes with shape [1]. 3. Adds support for unsigned types in an AtenItemOp folder 4. Adds a simpler quantized model for easier debugging 5. Adds a fusion pass to convert [quant -> dequant -> transpose -> mm] patterns to [transpose -> quant -> mm]. 6. Moved some xfails that are still not passing, but for different reasons than onnx.cast failures.
2024-04-02 07:21:05 +08:00
"QuantizedSingleLayer_basic",
Adds Some Quantization Support for AtenMatmulOp (#3147) 1. onnx.MatMulInteger now converts to aten.matmul instead of aten.mm 2. aten.matmul, for ranks >=2, now allows quantized inputs and will lower to linalg::quantized_matmul or linalg::quantized_batch_matmul. 3. added AtenMatmulOp to the FuseQuantizeOps rewrite patters QuantizeOperands, QuantizeTransposedOperands, and QuantizeAccumulator 4. added several tests, including some to test AtenMmOp with varying quantization signed-ness. 5. a quantized matmul mat-vec test is added to verify the failure to lower to linalg; cleaned of out-of-date code related to common torch-mlir lowering xfails. 6. in debugging a real model with quantized matmuls, I found a bug on the scalarize-shapes pass which resulted from the aten.full op folder returning an incompatible result type. This is fixed by the small change here to [lib/Dialect/Torch/IR/TorchOps.cpp](https://github.com/llvm/torch-mlir/compare/main...zjgarvey:torch-mlir:MatMulIntegerFix?expand=1#diff-dc8ed165c207918e606490eee3984b1ad51d7034e6aac36fc046bf47f6f03f4f).
2024-04-16 07:06:47 +08:00
"QuantizedBatchedInputSingleLayer_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
"SliceEndSleStartModule_basic",
"SliceOutOfUpperBoundIndexModule_basic",
Fixes for split tensor and slice (#2314) * RecomposeComplexOps: Remove dead slice op * lib/Dialect/Torch/IR/TorchOps.cpp: Fold slice ops even when they are on non-value tensors * lib/Conversion/TorchToTosa/TorchToTosa.cpp: Fix slice start/end out of range/none * lib/Dialect/Torch/IR/TorchOps.cpp: AtenSliceTensorOp::fold: Fold slices that go from 0:int_max * More tests for aten.split.Tensor
2023-07-20 15:53:54 +08:00
"SliceOutOfUpperBoundIndexStaticModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"SliceStartEqEndModule_basic",
"SqrtIntModule_basic",
"SubFloatModule_basic",
[torch] Supporting `torch.aten.mul.float` lowering to `arith` (#2833) Simple missing scalar operation for multiply floats was missing.
2024-02-06 08:23:04 +08:00
"MulFloatModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"SubIntModule_basic",
Support aten.stack op and decompose it into unsqueeze & cat (#1747)
2023-03-11 09:25:25 +08:00
"TensorsStackPromoteDTypeModule_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"TensorToIntZeroRank_basic",
"TensorToInt_basic",
"UniformModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
Add embedding Bag e2e case in xfail set (#1130)
2022-08-02 06:23:45 +08:00
"AtenEmbeddingBagSumExample_basic",
Add decomposition and E2E support for Aten_EmbeddingBag (#1137) * Add decomposition and E2E support for Aten_EmbeddingBag
2022-08-09 06:56:49 +08:00
"Aten_EmbeddingBagExample_basic",
Implement lowering of torch.aten.logit (#2697) Closes nod-ai/SHARK-Turbine#290
2024-01-11 22:55:42 +08:00
"ElementwiseLogitModule_basic",
E2E support for AtenRemainderScalarOp (#1200)
2022-08-11 08:02:06 +08:00
"ElementwiseRemainderScalarModule_Int_Float_basic",
"ElementwiseRemainderScalarModule_Bool_basic",
Implement lowering of torch.aten.lerp.Scalar (#2773) Closes nod-ai/SHARK-Turbine#356
2024-02-01 01:39:38 +08:00
"ElementwiseLerpScalarIntModule_basic",
"ElementwiseLerpScalarFloatModule_basic",
[MLIR][TORCH] Add Byte and Char Dtype support
2022-09-20 02:50:51 +08:00
"AtenIntTensorByteDtypeModule_basic",
"AtenIntTensorCharDtypeModule_basic",
[MLIR][TORCH] Add E2E support for aten.upsample_nearest2d_backward.vec op Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-11-01 21:08:04 +08:00
"UpSampleNearest2dBackwardVec_basic",
"UpSampleNearest2dBackwardOutputSizeNone_basic",
[MLIR][TORCH] Add E2E support aten.convolution_backward op This commit adds the decomposition for the `aten.convolution_backward` and `aten.convolution_backward_overrideable` op.
2022-11-04 15:57:29 +08:00
"ConvolutionBackwardModule2D_basic",
"ConvolutionBackwardModule2DPadded_basic",
[MLIR][TORCH] Add decomposition for aten.var_mean.correction op Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-11-15 22:39:40 +08:00
"VarMeanCorrectionModule_basic",
[MLIR][TORCH] Add E2E support for prims.convert_element_type op Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-11-21 16:38:47 +08:00
"VarMeanCorrectionNoneModule_basic",
Prelu lowering to linalg (#1712) Prelu lowering to linalg
2022-12-28 11:21:33 +08:00
"ElementwisePreluModule_basic",
[MLIR][TORCH] Add e2e support for aten.var_mean op Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-12-09 23:22:26 +08:00
"VarMeanBiasedModule_basic",
"VarMeanUnbiasedModule_basic",
[MLIR][TORCH] Add decomposition for aten.randn_like op This commit decomposes aten.randn_like op into aten.randn.generator op. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-01-16 19:40:21 +08:00
"RandnLikeModule_basic",
"RandnLikeDtypeModule_basic",
Decompose AtenNormalFunctionalOp into AtenRandn* and other arithmetic. (#2737)
2024-01-16 14:49:29 +08:00
"NormalFunctionalModule_basic",
build: manually update PyTorch version Set PyTorch and TorchVision version to nightly release 2023-02-27. This commit also adds the lowering for aten.add and aten.Float.Scalar op. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-02-28 10:32:21 +08:00
"BernoulliFloatModule_basic",
"BernoulliModule_basic",
"BernoulliPModule_basic",
"DropoutTrainModule_basic",
[Torch Dialect] Support aten.native_dropout (#2259) * [Torch Dialect] Support aten.native_dropout * update
2023-06-27 14:19:33 +08:00
"DropoutTrainStaticShapeModule_basic",
"NativeDropoutTrainModule_basic",
"NativeDropoutTrainStaticShapeModule_basic",
build: manually update PyTorch version Set PyTorch and TorchVision version to nightly release 2023-02-27. This commit also adds the lowering for aten.add and aten.Float.Scalar op. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-02-28 10:32:21 +08:00
"StdCorrectionKeepDimModule_basic",
"StdCorrectionNoneModule_basic",
"VarCorrectionKeepDimModule_basic",
"VarCorrectionNoneModule_basic",
[MLIR][TORCH] Add decomposition for prims::squeeze op This commit adds the decomposition for the prims.squeeze op. Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2022-11-16 13:57:58 +08:00
"AtenFloatScalarModule_basic",
"PrimsSqueezeModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
[MLIR][TORCH] Add support for prims::view_of op This op does nothing and just returns the input operand as the result of the op. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-04-10 11:50:26 +08:00
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
[Torch Dialect] support for aten.one_hot (#1852)
2023-04-11 16:02:28 +08:00
"OneHotModule_basic",
[MLIR][TORCH] Add E2E support for aten.var_mean.dim op This commit adds the decomposition for the aten.var_mean.dim op. Signed-Off By: Vivek Khandelwal<vivek@nod-labs.com>
2023-04-26 15:14:06 +08:00
"VarMeanDimModule_basic",
"VarMeanDimBiasedModule_basic",
Add complex types support with basic complex ops. Add complex types support with basic complex types. Add aten.imag and aten.real op lowering via linalg_backend.
2023-04-20 00:22:57 +08:00
"AtenComplexImagModule_basic",
"AtenComplexRealModule_basic",
[Torch Op] Add unbind.int support with ListUnpack (#2058) * add unbind int * reformat * use unpack canonicalize * address comments * Empty commit, trigger test * add ltc blacklist * clean up * address comments * check permute list * erase in recompose --------- Co-authored-by: zhekun.zhang <zhekun.zhang@bytedance.com>
2023-05-19 10:07:58 +08:00
"AtenComplexViewModule_basic",
[MLIR][TORCH] Add E2E support for view_as_real op (#2419) * view_as_real test case, allow dtype in testutils.randn * abstract python upstream func implemented * fixed upstream dtype func, implemented view_as_real backend op * formatted AtenViewAsRealOp, removed change in e2etest/framework * removed test suit from reshape_like.py, because it's moved to basic.py * implemented C-API wrapper for mlirComplexF128 type * fixed torch.complex dtype width in MLIR and Torch MLIR, deleted float16 dtype dict * Changed IR input of aten fft_fft unit test * code refactored * code refactored and fixed ci test * refactored: removed white spaces, and rolled back to having both input/output affine expr * refactored: deleted output affine expr to reduce redundancy * xfail ltc backend * removed ComplexImag and ComplexReal from torchdynamo xfail set * copied and pasted from main branch as there's no change to be made in this file * refactored abstract_interp_lib_gen.py * refactored: torchtypes.td, formatted, removed commented out code
2023-09-02 12:12:01 +08:00
"AtenRealView128Module_basic",
"AtenRealView64Module_basic",
[TM_TENSOR] Add `aten.scatter.[src|value]` op This commit adds support of `aten.scatter.src` and `aten.scatter.value` ops. Signed-Off-by: Gaurav Shukla <gaurav@nod-labs.com>
2022-10-16 05:46:06 +08:00
"ScatterValueFloatModule_basic",
"ScatterValueIntModule_basic",
[Torch Dialect] add folder for aten.add (#2264) * [Torch Dialect] add folder for aten.add * update * update * update
2023-06-27 10:55:28 +08:00
"UniformStaticShapeModule_basic",
[MLIR][TORCH] Add StableHLO lowering for embedding_bag.padding_idx op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-09-05 21:28:37 +08:00
"AtenEmbeddingBagStaticModule_basic",
[MLIR][TORCH] Add E2E support for aten.empty_strided decomposition op (redo PR) (#2459) Making the same PR with #2457, as I accidentally thought the review was already made and merged it (reverted). Add decompose empty_strided op. Referring to #1776, this decomposition op only supports default stride values, because accessing the tensor or indexing over that, the indices are determined by the strides. In MLIR, this is not implicitly supported but assumes that the strides are default while iterating over the tensor.
2023-09-14 01:04:31 +08:00
"EmptyStridedModule_basic",
Rework AtenEmptyStridedOp checks (#2537) Now using Value instead of Ints. Trades compile failure for a runtime assert
2023-11-01 11:56:54 +08:00
"EmptyStridedSizeIntStrideModule_basic",
[MLIR][TORCH] Add support for bitwise_right_shit and bitwise_and.Scalar op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-09-28 20:53:02 +08:00
"ElementwiseBitwiseAndScalarInt64Module_basic",
"ElementwiseBitwiseAndScalarInt32Module_basic",
[MLIR][TORCH] Add support for int8 dtype for sub, add, and bitwise_and op Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2023-10-03 19:59:56 +08:00
"ElementwiseBitwiseAndScalarInt8Module_basic",
[torch] Support `torch.convolution` quantized lowering to `linalg` (#2811) Linalg has quantized specific operations. We can lower to these operations when there is a known zeropoint and scale operations. This allows the `convolution` to occur with lower bitwidth's, improving the overall performance.
2024-01-31 05:46:47 +08:00
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[TorchToLinalg] Adds Quantization Support for ConvTranspose (#3240) I spent a little while debugging numerics issues with some tests similar to the ones in quantized_models.py, only to find that pytorch's quantized conv transpose is catastrophically inaccurate. I'll upstream the issue and only leave the tests here which are of the form quantize -> dequantize -> op.
2024-05-01 00:23:09 +08:00
"ConvTranspose2DQInt8_basic",
Added e2e LTC tests (#916) * Added e2e LTC Torch MLIR tests * Fix seed for reproducability * Check if computation is None before getting debug string * Updated unit tests, and added numeric tests * Print name of the model layer that fails numeric validation * Run LTC e2e test with CI/CD * Set seed in main function, instead of beginning of execution * Add comment to specify number of digits of precision * Fixed typo * Remove tests for LTC example models * Added LTC option to torchscript e2e * Implement compile and run for LTC e2e test * xfail all tests that use ops that aren't currently supported
2022-06-10 03:56:01 +08:00
}
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
ONNX_XFAIL_SET = {
Add per-test timeouts to catch infinite loops (#3650) Previously we only had full suite timeouts, making it impossible to identify which specific tests were hanging. This patch adds: 1. Per-test timeout support in the test framework 2. A default 600s timeout for all tests 3. A deliberately slow test to verify the timeout mechanism works The timeout is implemented using Python's signal module. Tests that exceed their timeout are marked as failures with an appropriate error message. This should help catch and isolate problematic tests that enter infinite loops, without needing to re-run the entire suite multiple times.
2024-08-22 02:37:31 +08:00
# This test is expected to time out
"TimeOutModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
# Failure - cast error
"PermuteNegativeIndexModule_basic",
# Failure - incorrect numerics
[Stablehlo] support aten.any.dim, aten.min.dim (#3500) * refactor `TorchToStablehlo/Reduction.cpp` * add `ConvertAtenReduceWithIndicesOp` patterns
2024-06-29 16:53:33 +08:00
"ReduceAnyDimFloatModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AvgPool2dDivisorOverrideModule_basic",
"BroadcastDynamicDimModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ElementwiseAtan2TensorIntModule_basic",
[Stablehlo] Support AtenPowScalarOp, AtenTanOp, AtenAsinhOp, AtenAcoshOp, AtenAtanhOp, Atan2Op (#3233)
2024-04-26 15:47:44 +08:00
"ElementwiseAtan2TensorIntStaticModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ElementwiseAtenFloorDivideScalarNegativeModule_basic",
"ElementwiseAtenFloorDivideTensorNegativeModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ElementwiseLog10IntModule_basic",
"ElementwiseLog2IntModule_basic",
[torch dialect] emit aten.fmax/fmin and add decomposition patterns (#3510)
2024-06-29 00:07:55 +08:00
"ElementwiseFminModule_basic",
"ElementwiseFmaxModule_basic",
[TorchToLinalg] address a dtype mismatch in `aten.multinomial` lowering (#3630) Resolves <https://github.com/llvm/torch-mlir/issues/3628> Unblocks a compile failure for one of the MiGraphx models (`AgentModel`).
2024-08-21 04:14:48 +08:00
"MultinomialModule2D_basic",
"MultinomialModule2D_F32",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"PixelShuffleModuleStaticRank4Float32_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ReflectionPad1dModule2dInput_Right",
"ReflectionPad1dModule2dInput_basic",
"ReflectionPad1dModule3dInput_Left",
"ReflectionPad1dModule3dInput_basic",
"ReflectionPad2dModule_Bottom",
"ReflectionPad2dModule_Left",
"ReflectionPad2dModule_Right",
"ReflectionPad2dModule_Top",
"ReflectionPad2dModule_basic",
"ReplicationPad2dModule_basic",
"ReplicationPad2dModule_bottom0",
"ReplicationPad2dModule_left0",
"ReplicationPad2dModule_right0",
"ReplicationPad2dModule_top0",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"SliceCopyEndGreaterThanDimSize_Module_basic",
"SliceCopyNegative_Module_basic",
"SliceCopyNonZeroDim_Module_basic",
"SliceCopy_Module_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"StdCorrectionLargeInputModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"TupleModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"VarCorrectionLargeInputModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
# Failure - incorrect shape
"ArangeStartOutDtypeModule_basic",
"ArangeStartOutViewModule_basic",
"MoveDimIntNegativeIndexModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ReduceL3NormKeepDimModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ViewSizeFromOtherTensor_basic",
[TorchToLinalg] address a dtype mismatch in `aten.multinomial` lowering (#3630) Resolves <https://github.com/llvm/torch-mlir/issues/3628> Unblocks a compile failure for one of the MiGraphx models (`AgentModel`).
2024-08-21 04:14:48 +08:00
# incorrect shape generated by torch.onnx.export (needs an unsqueeze)
"MultinomialModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
# Failure - onnx_export
"AdaptiveAvgPool1dGeneralDynamic_basic",
"AdaptiveAvgPool1dNonUnitOutputSizeDynamicModule_basic",
"AdaptiveAvgPool1dStaticLargerOutput_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AdaptiveAvgPool2dDynamicNoBatch_basic",
"AdaptiveAvgPool2dDynamic_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AdaptiveAvgPool2dNonUnitOutputSizeDynamicModule_basic",
[Torch] Add general AdaptiveAvgPool2dOp decompose support (#3111) Previously, it could only handle the situations where outputsize == (1, 1) or outputsize == (input_H, input_W). Now it supports all situations where input_H % output_H== 0 && input_W % output_W == 0
2024-04-11 17:02:59 +08:00
"AdaptiveAvgPool2dOutputSizeDivisibleByInputDynamicModule_basic",
[Torch] Add support for static uneven divisible AdaptiveAvgPool2d (#3566) The static uneven divisible AdaptiveAvgPool2d means that although the input size is not an integer multiple of ouput size, but the kernel and stride size can also be fixed (not dynamic). The derivation logic of kernel and stride size is consistent with torch/_decomp/decomposations.py:adaptive_avg_pool2d as described in the following: 1. Stride Size Firstly , derive the start index in each reduce operation according to the output size (`n`), `start_index = ([0, 1, ..., n - 1] * input_size) // output_size`. For each index `k`, if `k * (input_size % output_size) < output_size`, then the current and previous stride keeps the same as `input_size // output_size`. So suppose `(n-1) * (input_size % output_size) < output_size`, the stride in the whole AdaptiveAvgPool2d process keeps static, as `input_size // output_size`. 2. Kernel Size torch/_decomp/decomposations.py:adaptive_avg_pool2d calculates a static kernel size when the input/output sizes satisfy either of the two conditions, `input_size % output_size == 0` or `output_size % (input_size % output_size) == 0`. Here if `input_size % output_size == 0`, then the kernel size equals `input_size // output_size`, otherwise `input_size // output_size + 1.`
2024-08-01 11:37:53 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AdaptiveAvgPool3dDynamicNoBatch_basic",
"AdaptiveAvgPool3dDynamic_basic",
"AdaptiveMaxPool1dDynamicNoBatch_basic",
"AdaptiveMaxPool1dDynamic_basic",
"AdaptiveMaxPool1dStatic_basic",
"AdaptiveMaxPool2dDynamicNoBatch_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AdaptiveMaxPool2dDynamicWithIndices_basic",
"AdaptiveMaxPool2dDynamic_basic",
"AdaptiveMaxPool2dStaticWithIndices_basic",
"AdaptiveMaxPool2dStatic_basic",
Converts all Adaptive Pooling Ops to Linalg (#2808) The previous conversions for AtenAdaptiveAvgPool1dOp and AtenAdaptiveMaxPool2dOp are refactored into a general templated conversion that works for all of the AtenAdaptive...PoolNdOp's. New support is added for the following ops: 1. AtenAdaptiveMaxPool1d 2. AtenAdaptiveMaxPool3d 3. AtenAdaptiveAvgPool3d Support is also provided for passing inputs without batch dimensions. For example, applying adaptive_avg_pool2d to an input tensor of rank 3. After [pytorch #118162](https://github.com/pytorch/pytorch/pull/118162) gets down to torch-mlir, I'll add a test for AdaptiveMaxPool1d with return_indices (which will pass with that upstream fix). --------- Co-authored-by: James Newling <james.newling@gmail.com>
2024-03-23 02:05:20 +08:00
"AdaptiveMaxPool3dDynamicNoBatch_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AdaptiveMaxPool3dDynamicWithIndices_basic",
"AdaptiveMaxPool3dDynamic_basic",
"AdaptiveMaxPool3dStaticWithIndices_basic",
"AdaptiveMaxPool3dStatic_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AddCDivModule_basic",
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"Add_Module_basic",
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
"AtenComplex64Module_basic",
"AtenComplexImagModule_basic",
"AtenComplexRealModule_basic",
"AtenComplexViewModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AtenDiagEmbedDefaultDiag_basic",
"AtenDiagEmbedDimDiag_basic",
"AtenDiagEmbedNegOffsetDiag_basic",
"AtenDiagEmbedNonDefault4DDiag_basic",
"AtenDiagEmbedOffsetDiag_basic",
"AtenDiagEmbedRevDimDiag_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AtenEmbeddingBagStaticModule_basic",
"AtenEmbeddingBagSumExample_basic",
"AtenFloatScalarModule_basic",
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
"AtenIntTensorByteDtypeModule_basic",
"AtenIntTensorCharDtypeModule_basic",
[stablehlo] add aten::_int_mm lowering (#3474) as title
2024-06-20 16:10:31 +08:00
"AtenIntMM_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AtenItemFpOpModule_basic",
"AtenItemIntOpModule_basic",
Implement lowering of torch.aten.kthvalue (#3360) Closes [nod-ai/SHARK-Turbine#620](https://github.com/nod-ai/SHARK-Turbine/issues/620)
2024-06-15 13:48:39 +08:00
"AtenKthvalueModule_basic",
"AtenKthvalueKeepDimModule_basic",
"AtenKthvalueDynamicDimsModule_basic",
"AtenKthvalueFloat64Module_basic",
"AtenKthvalueFloat64DynamicDimsModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AtenLinalgCrossDynamic_basic",
Adds Some Quantization Support for AtenMatmulOp (#3147) 1. onnx.MatMulInteger now converts to aten.matmul instead of aten.mm 2. aten.matmul, for ranks >=2, now allows quantized inputs and will lower to linalg::quantized_matmul or linalg::quantized_batch_matmul. 3. added AtenMatmulOp to the FuseQuantizeOps rewrite patters QuantizeOperands, QuantizeTransposedOperands, and QuantizeAccumulator 4. added several tests, including some to test AtenMmOp with varying quantization signed-ness. 5. a quantized matmul mat-vec test is added to verify the failure to lower to linalg; cleaned of out-of-date code related to common torch-mlir lowering xfails. 6. in debugging a real model with quantized matmuls, I found a bug on the scalarize-shapes pass which resulted from the aten.full op folder returning an incompatible result type. This is fixed by the small change here to [lib/Dialect/Torch/IR/TorchOps.cpp](https://github.com/llvm/torch-mlir/compare/main...zjgarvey:torch-mlir:MatMulIntegerFix?expand=1#diff-dc8ed165c207918e606490eee3984b1ad51d7034e6aac36fc046bf47f6f03f4f).
2024-04-16 07:06:47 +08:00
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
[TorchToLinalg] Adds Support for Remaining Quantized Matmul Cases (#3167) The new cases added for quantized matmuls are: 1. vec-vec 2. vec-mat 3. mat-vec each of which are now lowered to expand(s), quantized_matmul, and collapse.
2024-04-17 00:28:28 +08:00
"AtenMatmulQint8VM_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AtenMatmulQint8VV_basic",
Adds Some Quantization Support for AtenMatmulOp (#3147) 1. onnx.MatMulInteger now converts to aten.matmul instead of aten.mm 2. aten.matmul, for ranks >=2, now allows quantized inputs and will lower to linalg::quantized_matmul or linalg::quantized_batch_matmul. 3. added AtenMatmulOp to the FuseQuantizeOps rewrite patters QuantizeOperands, QuantizeTransposedOperands, and QuantizeAccumulator 4. added several tests, including some to test AtenMmOp with varying quantization signed-ness. 5. a quantized matmul mat-vec test is added to verify the failure to lower to linalg; cleaned of out-of-date code related to common torch-mlir lowering xfails. 6. in debugging a real model with quantized matmuls, I found a bug on the scalarize-shapes pass which resulted from the aten.full op folder returning an incompatible result type. This is fixed by the small change here to [lib/Dialect/Torch/IR/TorchOps.cpp](https://github.com/llvm/torch-mlir/compare/main...zjgarvey:torch-mlir:MatMulIntegerFix?expand=1#diff-dc8ed165c207918e606490eee3984b1ad51d7034e6aac36fc046bf47f6f03f4f).
2024-04-16 07:06:47 +08:00
"AtenMatmulQint8_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"AtenMmQMixedSigni8_basic",
"AtenMmQint8_basic",
"AtenMmQuint8_basic",
[MLIR][TORCH] Add E2E support for aten.polar op (#3671) Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-09-03 13:21:03 +08:00
"AtenPolarFloatModule_basic",
"AtenPolarDoubleModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"AtenRealView128Module_basic",
"AtenRealView64Module_basic",
"AtenSubFloatModule_basic",
"AtenTopKModule_basic",
"AtenTopKSmallestModule_basic",
"Aten_EmbeddingBagExample_basic",
"AvgPool2dWithoutPadModule_basic",
"BatchMlpLayerModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"CeilFloatModule_basic",
"ChunkListUnpackDynamic_Module_basic",
"ChunkListUnpackUnevenDynamic_Module_basic",
"CollapseAllDimensionsModule_basic",
"CollapseFullDynamicModule_basic",
"CollapsePartialDynamicModule_basic",
"CollapseRank1DynamicModule_basic",
"CollapseStaticModule_basic",
"ConstantBoolParameterModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"Conv1dModule_basic",
"Conv2dBiasNoPaddingModule_basic",
"Conv2dModule_basic",
"Conv2dNoPaddingModule_basic",
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"Conv2dWithPaddingDilationStrideModule_basic",
"Conv2dWithPaddingModule_basic",
"Conv3dModule_basic",
"ConvTbcModule_basic",
[TorchToLinalg] Adds Quantization Support for ConvTranspose (#3240) I spent a little while debugging numerics issues with some tests similar to the ones in quantized_models.py, only to find that pytorch's quantized conv transpose is catastrophically inaccurate. I'll upstream the issue and only leave the tests here which are of the form quantize -> dequantize -> op.
2024-05-01 00:23:09 +08:00
"ConvTranspose2DQInt8_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"Conv_Transpose2dModule_basic",
"Convolution2DModule_basic",
"Convolution2DStridedModule_basic",
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2DStatic_basic",
"ConvolutionBackwardModule2DStrided_basic",
"ConvolutionBackwardModule2D_basic",
"ConvolutionModule2DGroups_basic",
"ConvolutionModule2DTransposeNonUnitOutputPadding_basic",
"ConvolutionModule2DTransposeStrided_basic",
"ConvolutionModule2DTranspose_basic",
"DivFloatModule_basic",
"DivIntModule_basic",
"ElementwiseAcoshIntModule_basic",
"ElementwiseAcoshModule_basic",
"ElementwiseAsinhIntModule_basic",
"ElementwiseAsinhModule_basic",
"ElementwiseAtanhIntModule_basic",
"ElementwiseAtanhModule_basic",
"ElementwiseAtenIsneginfOpModule_basic",
"ElementwiseAtenIsposinfOpModule_basic",
"ElementwiseBitwiseNotInt32Module_basic",
"ElementwiseBitwiseNotInt64Module_basic",
"ElementwiseBitwiseOrModule_basic",
"ElementwiseBitwiseOrStaticShapeModule_basic",
"ElementwiseBitwiseRightShiftInt32Module_basic",
"ElementwiseBitwiseRightShiftInt64Module_basic",
"ElementwiseBitwiseRightShiftInt8Module_basic",
"ElementwiseBitwiseXorModule_basic",
"ElementwiseBitwiseXorStaticShapeModule_basic",
[Stablehlo] support lowering sinh & cosh to stablehlo (#3213)
2024-04-23 19:54:58 +08:00
"ElementwiseSinhIntModule_basic",
"ElementwiseSinhModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseCoshIntModule_basic",
"ElementwiseCoshModule_basic",
[Torch] support aten.trunc (#3219) decompose `trunc(x)` to `sign(x) * floor(abs(x))`
2024-04-24 14:32:33 +08:00
"ElementwiseTruncIntModule_basic",
"ElementwiseTruncModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseDequantizePerChannelModule_basic",
"ElementwiseDequantizePerTensorModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ElementwiseDivTensorRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncIntStaticModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseEluNonDefaultModule_basic",
"ElementwiseExpm1IntModule_basic",
"ElementwiseExpm1Module_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ElementwiseFmodTensor_Int_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseMulTensorComplexModule_basic",
[linalg] Fix bug for conversion of complex dtype (#3269) The conversion of complex type wasn't supported or checked; the support and required tests were added. Fixes: https://github.com/iree-org/iree/issues/17226#issuecomment-2087779158
2024-05-01 14:36:53 +08:00
"ElementwiseMulTensorComplexDiffModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseOrTensorModule_basic",
"ElementwiseOrTensorStaticShapeModule_basic",
"ElementwiseQuantizePerTensorModule_basic",
[torch] Fix clamp ranges on quantize_per_tensor on unsigned (#3018) SExtValue was used for `int` and `uint` clamp values. This caused the result to always be outputed as `zero`.
2024-03-21 04:37:47 +08:00
"ElementwiseQuantizePerTensorUIntModule_basic",
Implement lowering for torch.aten.rad2deg (#3586)
2024-08-14 19:07:28 +08:00
"ElementwiseRad2DegModule_basic",
"ElementwiseRad2DegIntModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ElementwiseRemainderTensorModule_Int_basic",
[TorchToLinalg] Fix torch.aten.remainder for negative operands (#3581) Closes #3575 The PyTorch remainder operator is meant to compute the Python modulus operator entrywise: https://pytorch.org/docs/stable/generated/torch.remainder.html#torch.remainder In python the modulus operator is meant to always return a result with the same sign as the divisor: https://docs.python.org/3/reference/expressions.html#binary-arithmetic-operations In other words, torch.aten.remainder should return a Python-style modulus instead of a C-style modulus. However the remainder operator was simply translated into arith.ModSI or arith.ModF, which both effectively compute the C-style modulus. Now the lowering has been modified so that the modulus operator works properly with negative numbers, both in the dividend, and the divisor.
2024-08-13 23:47:21 +08:00
"ElementwiseRemainderTensorModule_Int_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDivisor_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ElementwiseSgnModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"EmptyStridedModule_basic",
"EmptyStridedSizeIntStrideModule_basic",
"EqIntModule_basic",
"ExponentialModule_basic",
Support fake_quantize_per_tensor_affine_cachemask (#3477) Add a new op with shape/dtypes and decompose into `fake_quantize_per_tensor_affine` when the second result is unused. The xfail_set change is on ONNX because torch cannot export this op to ONNX.
2024-06-21 15:15:31 +08:00
"FakeQuantizePerTensorAffineCachemaskModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"FloatImplicitModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
"GeluBackwardModule_basic",
"GtFloatIntModule_basic",
"GtIntModule_basic",
"HardtanhBackward_basic",
Implement lowering of torch.aten.hstack (#3563)
2024-09-11 19:11:47 +08:00
"HstackBasicComplexModule_basic",
"HstackBasicFloatModule_basic",
"HstackBasicIntFloatModule_basic",
"HstackBasicIntModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"IndexPutImpl1DFloatAccumulateModule_basic",
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
"IndexPutImpl2DFloatAccumulateModule_basic",
"IndexPutImpl2DFloatNonAccumulateModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"IndexPutImpl2DImplicitModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"IndexPutImpl2DIndexModule_basic",
"IndexPutImpl2DNoneIndexStaticModule_basic",
"IndexPutImpl3DFloatAccumulateModule_basic",
"IndexPutImpl3DFloatNonAccumulateModule_basic",
"IndexPutImplIndexWithNoneModule_basic",
[LINALG] Broadcast `values` to shape of slize in `index_put` (#3487) The `index_put` operation, `input[indices] = values`, allows for the values to be any shape that is broadcastable to the slice `input[indices]`. This commit adds broadcasting support to the Linalg lowering of `IndexPutHackedTwinOp`. Fixes: #3465
2024-06-26 16:59:49 +08:00
"IndexPutWithNoneAndBroadcastModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"IntFloatModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"IntImplicitModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"IouOfModule_basic",
"IsFloatingPointFloat_True",
"IsFloatingPointInt_False",
"IscloseStaticModuleTrue_basic",
"IscloseStaticModule_basic",
"LeakyReluBackwardModule_basic",
"LeakyReluBackwardStaticModule_basic",
"LenStrModule_basic",
"LiftFreshCopyModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"LinalgNormKeepDimComplexModule_basic",
"LinalgVectorNormComplexModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"LogSoftmaxBackwardModule_basic",
[Torch] Decompose AtenMaskedScatterOp (#3353) Co-authored-by: Yuanqiang Liu <liuyuanqiang.yqliu@bytedance.com>
2024-05-16 15:27:25 +08:00
"MaskedScatterStaticBasic_basic",
MaxPool1d lowering to linalg (#3295) Co-authored-by: root <root@i32b01216.sqa.eu95>
2024-05-11 00:35:26 +08:00
"MaxPool1dCeilModeTrueModule_basic",
"MaxPool1dModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"MaxPool2dCeilModeTrueModule_basic",
"MaxPool2dModule_basic",
"MaxPool2dWithIndicesAllOnesModule_basic",
"MaxPool2dWithIndicesBackwardDynamic3DModule_basic",
"MaxPool2dWithIndicesBackwardDynamic4DModule_basic",
"MaxPool2dWithIndicesBackwardStatic3DModule_basic",
"MaxPool2dWithIndicesBackwardStatic4DModule_basic",
"MaxPool2dWithIndicesCeilModeTrueModule_basic",
"MaxPool2dWithIndicesFullSizeKernelModule_basic",
"MaxPool2dWithIndicesModule_basic",
"MaxPool2dWithIndicesNonDefaultDilationModule_basic",
"MaxPool2dWithIndicesNonDefaultParamsModule_basic",
"MaxPool2dWithIndicesNonDefaultStrideModule_basic",
"MaxPool3dCeilModeTrueModule_basic",
"MaxPool3dLargeDatadModule_basic",
"MaxPool3dModuleRandomSimple_basic",
"MaxPool3dModule_basic",
[TorchToLinalg] Support lowering MaxPool3dWithIndices (#3652) Support torch.MaxPool3dWithIndices lowering to linalg backend.
2024-08-28 03:14:25 +08:00
"MaxPool3dWithIndicesAllOnesModule_basic",
"MaxPool3dWithIndicesCeilModeTrueModule_basic",
"MaxPool3dWithIndicesFullSizeKernelModule_basic",
"MaxPool3dWithIndicesModule_basic",
"MaxPool3dWithIndicesNonDefaultDilationModule_basic",
"MaxPool3dWithIndicesNonDefaultParamsModule_basic",
"MaxPool3dWithIndicesNonDefaultStrideModule_basic",
`max_unpool3d` linalg lowering (#3536) An attempt of `aten.max_unpool3d` to linalg lowering. There are known issues with this implementation (see comment in code).
2024-07-31 01:59:17 +08:00
"MaxUnpool3dModule_basic",
"MaxUnpool3dModulePad0_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"MeanDimEmptyDimModule_basic",
"Mlp1LayerModule_basic",
"Mlp2LayerModuleNoBias_basic",
"Mlp2LayerModule_basic",
"MulFloatModule_basic",
"MulIntModule_basic",
"NarrowHorizontalTest2_basic",
"NarrowHorizontalTest_basic",
"NarrowTensorHorizontalModule_basic",
"NarrowTensorVerticalModule_basic",
"NarrowVerticalTest2_basic",
"NarrowVerticalTest_basic",
"NativeBatchNorm1DModule_basic",
"NativeBatchNorm2DModule_basic",
"NativeBatchNorm3DModule_basic",
"NativeBatchNormNoneWeightModule_basic",
"NativeDropoutEvalFloatModule_basic",
"NativeGroupNormBackwardModule_basic",
"NativeGroupNormModule_basic",
"NativeLayerNormDynamicModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
"NewEmptyStridedModuleDefaultDtype_basic",
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
"NllLossModuleBackwardMeanWeight_basic",
"NllLossModuleBackwardMean_basic",
"NllLossModuleBackwardSumWeight_basic",
"NllLossModuleBackwardSum_basic",
"NllLossModuleBackwardWeight_basic",
"NllLossModuleBackward_basic",
"NllLossModuleBackward_ignore_index",
"NllLossModule_1D_basic",
"NllLossModule_basic",
"NllLossModule_ignore_index_out_of_bounds_basic",
"NllLossModule_mean_basic",
"NllLossModule_sum_basic",
Add complex support for aten.norm and similar operations (#3052) Add support for complex-type input tensors for norm, vector norm, and Frobenius norm operations.
2024-04-02 16:33:30 +08:00
"NormScalarComplexModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"NormScalarModule_basic",
"NormScalarOptDimKeepDimComplexModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"NormScalarOptDimKeepDimModule_basic",
"NormScalarOptDimModule_basic",
"NormalFunctionalModule_basic",
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"PixelShuffleModuleFullDynamic_basic",
"PixelShuffleModuleSpatiallyDynamic_basic",
"PixelShuffleModuleSpatiallyStatic_basic",
"PixelShuffleModuleStaticRank3Int64_basic",
"PowIntFloatModule_basic",
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsConvertElementTypeModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
[torch] Adds Quantization Support for `aten.relu` (#3177) A choice was made to quantize the return type of Relu with a scale and zero point copied from the input's quantization scheme. With this choice, the torch-to-linalg conversion of quantized Relu essentially computes max(input, zeroPoint) in the elementwise payload.
2024-04-24 02:01:36 +08:00
"QuantizedReluInt8_basic",
"QuantizedReluInt32_basic",
"QuantizedReluUint8_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"RandIntDtypeModule_basic",
"RandIntModule_basic",
"RandIntPinMemoryModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ReduceFrobeniusNormComplexModule_basic",
"ReduceL1NormComplexModule_basic",
"ReduceL2NormComplexModule_basic",
"ReduceL3NormKeepDimComplexModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ReshapeAliasCollapseModule_basic",
"ReshapeAliasExpandModule_basic",
"ReshapeExpandModule_basic",
[LINALG] Implement lowering of torch.aten.rot90 (#3551)
2024-09-06 13:06:17 +08:00
"Rot90DynamicDimsModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ScalarConstantTupleModule_basic",
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
"ScatterReduceFloatMaxModule",
"ScatterReduceFloatMeanModule",
"ScatterReduceFloatMeanModuleIncludeSelf",
"ScatterReduceFloatMinModule",
"ScatterReduceFloatProdModule",
"ScatterReduceFloatSumModule",
"ScatterReduceIntMaxModule",
"ScatterReduceIntMeanModule",
"ScatterReduceIntMeanModuleIncludeSelf",
"ScatterReduceIntMinModule",
"ScatterReduceIntProdModule",
"ScatterReduceIntSumModule",
"SelectScattertModule_basic",
"SelectScattertStaticModule_basic",
implement lowering of torch.aten._linalg_slogdet (#3524)
2024-07-19 13:54:43 +08:00
"SignAndLogarithmOfDeterminantModule_F32",
"SignAndLogarithmOfDeterminantBatchedModule_F32",
"SignAndLogarithmOfDeterminantDynamicModule_F32",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"SliceEndSleStartModule_basic",
"SliceOutOfUpperBoundIndexModule_basic",
"SliceScatterModule_basic",
"SliceScatterNegativeDimModule_basic",
"SliceScatterNegativeEndModule_basic",
"SliceScatterStaticModule_basic",
"SliceScatterStepVariationModule_basic",
"SliceScatterZeroDimModule_basic",
"SliceStartEqEndModule_basic",
"SoftmaxBackwardModule_basic",
"SortIntListReverse_basic",
"SortIntList_basic",
"SplitDimDynamicModule_basic",
"SplitDimStaticModule_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SubFloatModule_basic",
"SubIntModule_basic",
"TanhBackward_basic",
[LINALG] Added support for conversion from float to complex. (#3595)
2024-08-07 15:06:48 +08:00
"TensorsConcatComplex128FloatModule_basic",
"TensorsConcatComplex64FloatModule_basic",
Added support for integer to complex conversion (#3604)
2024-08-14 20:43:00 +08:00
"TensorsConcatComplex128IntModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"TensorToIntZeroRank_basic",
"TensorToInt_basic",
"TestMultipleTensorAndPrimitiveTypesReturn_basic",
"Threshold1dFloatModule_basic",
"Threshold1dIntI32Module_basic",
"Threshold1dIntModule_basic",
"Threshold2dFloatModule_basic",
"Threshold2dIntModule_basic",
"Threshold3dFloatModule_basic",
"Threshold3dIntModule_basic",
"ThresholdBackward1dFloatModule_basic",
"ThresholdBackward1dIntModule_basic",
"ThresholdBackward1dMixedModule_basic",
"ThresholdBackward2dFloatModule_basic",
"ThresholdBackward2dIntModule_basic",
"ThresholdBackward2dMixedModule_basic",
"ThresholdBackward3dFloatModule_basic",
"ThresholdBackward3dIntModule_basic",
"ThresholdBackward3dMixedModule_basic",
"ToCopyBoolDTypeStaticModule_basic",
"ToCopyModule_basic",
"ToCopyWithDTypeFalsePinMemoryModule_basic",
"ToCopyWithDTypeModule_basic",
"TorchPrimLoopForLikeModule_basic",
"TorchPrimLoopWhileLikeModule_basic",
"TraceModule_basic",
"TraceModule_empty",
"TraceModule_nonsquare",
"TraceSignedIntModule_basic",
"TraceUnsignedIntModule_basic",
"TraceUnsignedIntModule_empty",
"UniformModule_basic",
"UniformNoCorrelationModule_basic",
"UniformStaticShapeModule_basic",
"UnsafeIndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"UnsafeView1DFoldModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"UnsafeViewCollapseModule_basic",
"UnsafeViewDynamicExpandModule_basic",
"UnsafeViewDynamicExpandWithAtenSizeIntModule_basic",
"UnsafeViewExpandModule_basic",
"UpSampleNearest2dBackwardScalesNone_basic",
"UpSampleNearest2dBackward_basic",
"UpSampleNearest2dDynamicFactor_basic",
"UpSampleNearest2dStaticFactor_basic",
"UpSampleNearest2d_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewCollapseModule_basic",
"ViewDynamicExpandCollapseModule_basic",
"ViewDynamicExpandCollapseWithAtenIntModule_basic",
"ViewDynamicExpandModule_basic",
"ViewDynamicExpandWithAtenSizeIntModule_basic",
"ViewExpandDynamicDimModule_basic",
"ViewNoChange1dModule_basic",
"ViewNoChange2dModule_basic",
"ViewNoChange3dModule_basic",
[Torch] Implements TorchToLinalg lowering of torch.ops.aten._weight_norm_interface (#3538) Resolves https://github.com/nod-ai/SHARK-Turbine/issues/757. Adds TorchToLinalg lowering for `Aten_WeightNormInterfaceOp`. --------- Co-authored-by: Ubuntu <rbhowmik@RohanBhowmikVM.judsoscro3wupi0qm4bjlj5m3b.bx.internal.cloudapp.net>
2024-07-17 01:39:12 +08:00
"WeightNormInterfaceModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"_Convolution2DAllFalseModule_basic",
"_Convolution2DBenchmarkModule_basic",
"_Convolution2DCudnnModule_basic",
"_Convolution2DDeterministicModule_basic",
"_Convolution2DTF32Module_basic",
"_ConvolutionDeprecated2DAllFalseModule_basic",
"_ConvolutionDeprecated2DBenchmarkModule_basic",
"_ConvolutionDeprecated2DCudnnModule_basic",
"_ConvolutionDeprecated2DDeterministicModule_basic",
"_SoftmaxModule_basic",
Support onnx.If (#2825) This is probably a decent PR for learning about blocks and regions. If you're here to learn about that, consider also looking at lib/Conversion/TorchToSCF/TorchToSCF.cpp While this doesn't include an e2e test, it is tested downstream in https://github.com/nod-ai/SHARK-TestSuite/blob/main/e2eshark/onnx/operators/If/model.py --------- Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
2024-05-01 02:36:40 +08:00
# Failure - onnx_import
[Torch] support recompose of aten.split.with_sizes and aten.tensor_sp… (#3401) …lit.sections * support recompose to aten.split.with_sizes and aten.tensor_split.sections * fix recompose of aten.chunk
2024-05-31 09:56:47 +08:00
# Failure - onnx_lowering: onnx.SplitToSequence
"ChunkListUnpackUneven_Module_basic",
"TensorSplitSections_GetItemModule_basic",
"TensorSplitSections_ListUnpackModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
# Failure - onnx_lowering: onnx.AveragePool
"AdaptiveAvgPool1dGeneralDynamicNoBatches_basic",
Support onnx.If (#2825) This is probably a decent PR for learning about blocks and regions. If you're here to learn about that, consider also looking at lib/Conversion/TorchToSCF/TorchToSCF.cpp While this doesn't include an e2e test, it is tested downstream in https://github.com/nod-ai/SHARK-TestSuite/blob/main/e2eshark/onnx/operators/If/model.py --------- Co-authored-by: Xida Ren <xida.ren.dev@gmail.com>
2024-05-01 02:36:40 +08:00
# these diagonal modules are currently failing due to dynamic shape.
# We are currently testing aten.diagonal using DiagonalWithStaticShapeModule instead.
# when the issue is fixed, please remove DiagonalWithStaticShapeModule as well as the xfails here.
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"TileBigDimsSizeModule_basic",
"TileSmallDimsSizeModule_basic",
[Torch] Support AtenProdOp on linalg and stablehlo (#3215)
2024-04-24 11:14:04 +08:00
# Failure - onnx_lowering: onnx.ReduceProd
"ReduceProdFloatModule_basic",
"ReduceProdDtypeFloatModule_basic",
"ReduceProdElementTypeBoolModule_basic",
"ReduceProdUnsignedIntModule_basic",
"ReduceProdSignedIntModule_basic",
"ReduceProdDtypeIntModule_basic",
[onnx] Add `onnx-to-torch` lowering for random ops (#3193) This commit adds the OnnxToTorch lowering for Onnx's RandomNormal, RandomNormalLike, RandomUniform, and RandomUniformLike op.
2024-04-23 00:58:07 +08:00
# ERROR: dtype (torch.float32) is not equal to golden dtype (torch.float64)
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"RandnDtypeDeviceModule_basic",
"RandnGeneratorF64Module_basic",
"RandnGeneratorModule_basic",
"RandnModule_basic",
"RandnLikeModule_basic",
"BernoulliFloatModule_basic",
"BernoulliPModule_basic",
"BernoulliTensorModule_basic",
# Failure - onnx_lowering: onnx.ReduceProd
"ReduceProdDimIntFloatModule_basic",
# Failure - onnx_lowering: onnx.ScatterElements
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ScatterReduceFloatMaxModuleIncludeSelf",
"ScatterReduceFloatMinModuleIncludeSelf",
"ScatterReduceIntMaxModuleIncludeSelf",
"ScatterReduceIntMinModuleIncludeSelf",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ScatterValueFloatModule_basic",
[Torch] add support for aten.scatter_add (#3534)
2024-07-12 09:15:42 +08:00
"ScatterAddStaticModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
# Failure - onnx_lowering: onnx.ScatterND
"IndexPut1DFloatAccumulateModule_basic",
"IndexPut1DIntAccumulateModule_basic",
"IndexPut2DFloatAccumulateModule_basic",
"IndexPut2DIntAccumulateModule_basic",
"IndexPut3DFloatAccumulateModule_basic",
"IndexPut3DIntAccumulateModule_basic",
"IndexPutHackedTwin1DFloatAccumulateModule_basic",
"IndexPutHackedTwin1DIntAccumulateModule_basic",
"IndexPutHackedTwin2DFloatAccumulateModule_basic",
"IndexPutHackedTwin2DIntAccumulateModule_basic",
"IndexPutHackedTwin3DFloatAccumulateModule_basic",
"IndexPutHackedTwin3DIntAccumulateModule_basic",
[Torch] Emit and decompose prims.iota op (#3132)
2024-04-22 10:45:01 +08:00
# RuntimeError: unsupported input type: Device
"PrimsIotaModule_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
# unimplemented torchvision.deform_conv2d torch->linalg
"DeformConv2D_basic",
Implement lowering of torch.aten.renorm (#3388) Closes [nod-ai/SHARK-Turbine/issues/689](https://github.com/nod-ai/SHARK-Turbine/issues/689) --------- Co-authored-by: Branko Trifkovic <branko.trifkovic@syrmia.com>
2024-06-18 01:40:57 +08:00
# Error: 'aten::renorm' to ONNX opset version 17 is not supported.
"RenormModuleFloat16_basic",
"RenormModuleFloat32NegativeDim_basic",
"RenormModuleFloat32_basic",
"RenormModuleFloat32DynamicDims_basic",
[LINALG] Implement lowering of torch.aten.rot90 (#3551)
2024-09-06 13:06:17 +08:00
"Rot90BasicModule_basic",
"Rot90DynamicDymsModule_basic",
"Rot90MultipleRotationsModule_basic",
"Rot90NegativeEvenRotationsModule_basic",
"Rot90NegativeOddRotationsModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
# Failure - unknown
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"BernoulliModule_basic",
[Torch]Support conv_transpose1d and conv_transpose3d (#3286) 1. Support conv_transpose1d and conv_transpose3d 2. Fix bugs of convertTransposedConv func in lib/Conversion/TorchToStablehlo/Linear.cpp
2024-06-03 15:11:12 +08:00
"Conv_Transpose1dModule_basic",
"Conv_Transpose3dModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"CopyWithDifferentDTypesAndSizesModule_basic",
"CopyWithDifferentDTypesModule_basic",
"CosineSimilarityStaticBroadcastModule_basic",
"CumsumInputDtypeInt32Module_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ElementwiseAcosIntModule_basic",
"ElementwiseAsinIntModule_basic",
"ElementwiseAtanTensorIntModule_basic",
"ElementwiseCosIntModule_basic",
Added 2 Ops: Floor divide scalar and Floor divide scalar mode (#3156) - Added linalg lowering for `AtenFloorDivideScalarOp` - Needed `AtenDivScalarModeOp` for the decomp. - Added linalg lowering for `AtenDivScalarModeOp` - Moved linalg payload logic to `createDivModePayload()` since the logic was nearly identical for both `AtenDivScalarModeOp` and `AtenDivTensorModeOp`. Just a template function - Added `AtenDivScalarModeOp` lowering for stablehlo Pytorch's [`torch.floor_divide()`](https://pytorch.org/docs/stable/generated/torch.floor_divide.html) in a previous version (for a reason unknown to me) preformed a truncation instead of "floor". The already implemented op `AtenFloorDivideTensorOp` was done before this change. However, this wasn't caught because our testcases only tested positive floor division. I changed this to floor as well as adding a few test cases.
2024-04-16 04:45:10 +08:00
"ElementwiseDivTensorRoundingModeTruncModule_basic",
"ElementwiseDivTensorRoundingModeTruncStaticModule_basic",
"ElementwiseErfIntModule_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ElementwiseExpIntModule_basic",
"ElementwiseLogIntModule_basic",
"ElementwiseSigmoidIntModule_basic",
"ElementwiseSinIntModule_basic",
"ElementwiseTanIntModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ElementwiseToDtypeI64ToUI8Module_basic",
[onnx] Fix ReduceMean lowering to torch (#2956) Torch lowering only supported the most recent version. Refactored the lowering so more easily handle default values and optional operands / attributes.
2024-02-28 14:48:07 +08:00
"ElementwiseUnaryIntModule_basic",
[Linalg] Promote type for compare tensor op (#3416)
2024-06-05 07:05:39 +08:00
"ElementwiseFloatTensorGtIntTensorModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"MaskedFillTensorFloatValueModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"NativeDropoutTrainModule_basic",
"NativeDropoutTrainStaticShapeModule_basic",
[stablehlo] Support aten.any and aten.all lowering (#3217)
2024-04-25 11:15:52 +08:00
"ReduceAnyFloatModule_basic",
[onnx] Update the failure triage for onnx (#3186) Reclassifying what the source of failures are for various bugs so we can reprioritize what failures are common.
2024-04-19 05:58:13 +08:00
"ReduceMaxAlongDimUnsignedInt_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
"ReduceMinAlongDimUnsignedInt_basic",
[TOSA] Fix Tensor.hacked_twin to support diff size indexes (#3547) - Broadcasts index list tensors - Adds torch.nn.Unfold test Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-07-31 05:32:05 +08:00
"UnfoldModule_basic",
[onnx] Add testing using the `onnx` compilation using torch tests (#2795) We can route the torch tests via `onnx` using the `torch.onnx.export` tooling. We can then reimport, lower to torch, and compile to linalg to validate the onnx path is working correctly. The current implementation exposes some failures in the `onnx` path so we cannot enable the onnx test suite yet due to segmentation faults.
2024-02-16 02:17:13 +08:00
}
[onnx] Update the importer to create a `none` for missing operands (#2931) Some operands are optional so we require a placeholder for missing operands. We invent an `onnx.None` operation as our placeholder.
2024-02-21 01:30:30 +08:00
[NFC reformat] Applies pre-commit formatting to Python files. (#3244) This is a large change because prior to this point, Python files in the project were not consistently formatted. This reformats them all with black defaults. Based on experience with prior projects, if you have a dev/long-term branch with Python patches, you can minimize merge conflicts prior to rebasing to include this commit by running `black` on your modified Python files, squashing, and then rebasing/merging.
2024-04-28 05:16:31 +08:00
if torch_version_for_comparison() < version.parse("2.3.0.dev"):
[Torch] Add decomposition of RepeatInterleaveSelfInt Op (#3075) Decomposition RepeatInterleaveSelfInt with following ops: ```python def my_repeat_interleave(input, repeats, dim=None): if dim is None: # Flatten the input and then repeat return input.flatten().unsqueeze(-1).tile((1, repeats)).flatten() else: # Calculate the shape after repeat expanded_shape = list(input.shape) expanded_shape[dim] *= repeats # Repeat the tensor along the specified dimension repeat_shape = [1] * (input.dim() + 1) repeat_shape[dim + 1] = repeats input = input.unsqueeze(-1) # Tile and then reshape tiled = torch.tile(input, repeat_shape) # Rearrange and reshape repeated = tiled.reshape(*expanded_shape) return repeated ``` I passed the tests of stablehlo and linalg. When testing onnx, strange things happened. In torch-mlir's CI **torch_nightly** and my own environment(torch==2.4.0.dev20240318+cpu), it can **pass the pass**. In torch-mlir's CI **torch_stable**, it **failed**. The test case is `RepeatInterleaveSelfIntNoDimModule_basic`, the result shape should be [120]. ```python class RepeatInterleaveSelfIntNoDimModule(torch.nn.Module): def __init__(self): super().__init__() @export @annotate_args([ None, ([3, 4, 5], torch.float32, True), ]) def forward(self, x): return x.repeat_interleave(2) @register_test_case(module_factory=lambda: RepeatInterleaveSelfIntNoDimModule()) def RepeatInterleaveSelfIntNoDimModule_basic(module, tu: TestUtils): module.forward(tu.rand(3, 4, 5)) ``` The error log is as follows: ``` Unexpected outcome summary: (onnx) ****** Failed tests - 1 tests FAIL - "RepeatInterleaveSelfIntNoDimModule_basic" @ trace item #0 - call to "forward" @ output of call to "forward" ERROR: shape (torch.Size([6, 4, 5])) is not equal to golden shape (torch.Size([120])) ``` @rsuderman Would you please help me check what's wrong with my PR? Thanks a lot.
2024-04-18 06:27:51 +08:00
ONNX_XFAIL_SET = ONNX_XFAIL_SET | {
# ERROR: shape (torch.Size([6, 4, 5])) is not equal to golden shape (torch.Size([120]))
"RepeatInterleaveSelfIntNoDimModule_basic",
}
build: manually update PyTorch version (#3340) Set PyTorch and TorchVision version to nightly release 2024-05-14. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-06-07 00:53:40 +08:00
if torch_version_for_comparison() < version.parse("2.4.0.dev"):
ONNX_XFAIL_SET = ONNX_XFAIL_SET | {
# torch.onnx.errors.UnsupportedOperatorError: Exporting the operator 'aten::bitwise_left_shift' to ONNX opset version 17 is not supported.
"ElementwiseBitwiseLeftShiftInt32Module_basic",
"ElementwiseBitwiseLeftShiftInt64Module_basic",
"ElementwiseBitwiseLeftShiftInt8Module_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
# bitwise and support has been added in torch nightly
"ElementwiseAndScalarModule_basic",
"ElementwiseAndScalarStaticShapeModule_basic",
"ElementwiseBitwiseAndModule_basic",
"ElementwiseBitwiseAndScalarInt32Module_basic",
"ElementwiseBitwiseAndScalarInt64Module_basic",
"ElementwiseBitwiseAndScalarInt8Module_basic",
"ElementwiseBitwiseAndStaticShapeModule_basic",
build: manually update PyTorch version (#3340) Set PyTorch and TorchVision version to nightly release 2024-05-14. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-06-07 00:53:40 +08:00
}
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
if torch_version_for_comparison() >= version.parse("2.5.0.dev"):
ONNX_XFAIL_SET = ONNX_XFAIL_SET | {
# ERROR: value (Tensor with shape=[2, 3, 8, 20], dtype=torch.float32, min=+nan, max=+nan, mean=+nan) is not close to golden value (Tensor with shape=[2, 3, 8, 20], dtype=torch.float32, min=-2.394, max=+2.454, mean=-0.02828)
"ScaledDotProductAttentionBoolMaskModule_basic",
}
[stablehlo] add aten::_int_mm lowering (#3474) as title
2024-06-20 16:10:31 +08:00
if torch_version_for_comparison() < version.parse("2.4.0.dev"):
STABLEHLO_PASS_SET = STABLEHLO_PASS_SET - {
"AtenIntMM_basic",
}
FX_IMPORTER_STABLEHLO_XFAIL_SET = FX_IMPORTER_STABLEHLO_XFAIL_SET | {
"AtenIntMM_basic",
}
[MLIR][TORCH] Add OnnxToTorch lowering for ReduceL1 Op (#3146) Adds OnnxToTorch Lowering for the ReduceL1 op.
2024-04-16 14:54:46 +08:00
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
ONNX_CRASHING_SET = LINALG_CRASHING_SET | {
[Torch Dialect] support aten.fake_quantize_per_tensor_affine (#3014)
2024-03-15 08:53:29 +08:00
"FakeQuantizePerTensorAffineModule_basic",
"FakeQuantizePerTensorAffineDynamicShapeModule_basic",
[MLIR][TORCH] Fix OnnxToLinalg lowering issue for Squeeze and Unsqueeze op (#2991) This commit also cleans up the OnnxToTorch lowering for the Squeeze and Unsqueeze op and adds the support for handling edge cases. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-04-22 16:52:42 +08:00
"ElementwisePreluModule_basic",
[MLIR][TORCH] Support parallel dimemsions expand/collapse (#3051) This PR support `aten.view` with unique unknown dimension both in input shape and output shape while the pass convert-torch-to-linalg that lowing `aten.view` to `tensor.collapse_shape` or `tensor.expand_shape`. Below is an example ``` func.func @test_reshape(%arg0: !torch.vtensor<[1,?,50,16],f32>) -> !torch.vtensor<[1,?,16],f32> attributes {torch.assume_strict_symbolic_shapes, torch.onnx_meta.ir_version = 9 : si64, torch.onnx_meta.opset_version = 19 : si64, torch.onnx_meta.producer_name = "backend-test", torch.onnx_meta.producer_version = ""} { %int1 = torch.constant.int 1 %int-1 = torch.constant.int -1 %int16 = torch.constant.int 16 %0 = torch.prim.ListConstruct %int1, %int-1, %int16 : (!torch.int, !torch.int, !torch.int) -> !torch.list<int> %1 = torch.aten.view %arg0, %0 : !torch.vtensor<[1,?,50,16],f32>, !torch.list<int> -> !torch.vtensor<[1,?,16],f32> return %1 : !torch.vtensor<[1,?,16],f32> } ```
2024-04-12 01:43:03 +08:00
"ViewDynamicExpandCollapseWithParallelUnknownDimModule_basic",
[MLIR][TORCH] Fix OnnxToLinalg lowering issue for Squeeze and Unsqueeze op (#2991) This commit also cleans up the OnnxToTorch lowering for the Squeeze and Unsqueeze op and adds the support for handling edge cases. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-04-22 16:52:42 +08:00
"ScatterReduceFloatProdModuleIncludeSelf",
"ScatterReduceFloatSumModuleIncludeSelf",
"ScatterReduceIntProdModuleIncludeSelf",
"ScatterReduceIntSumModuleIncludeSelf",
Temporarily remove QuantizedMLP_basic (#3301) See issue #3298
2024-05-08 05:32:13 +08:00
# The following test sporadically stopped producing correct numerics for the golden value in the CI.
# For now, we are removing the test until this issue has been debugged.
"QuantizedMLP_basic",
[torch] Fix `onnx.MaxPool` lowering (#3133) This commit fixes the onnx.MaxPool op lowering which was lacking the indices result support. Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-05-21 23:35:32 +08:00
# Runtime crash: mismatched size for broadcast
"MaxPool2dWithIndicesAllNegativeValuesModule_basic",
"MaxPool2dWithIndicesNonDefaultPaddingModule_basic",
[TorchToLinalg] Support lowering MaxPool3dWithIndices (#3652) Support torch.MaxPool3dWithIndices lowering to linalg backend.
2024-08-28 03:14:25 +08:00
"MaxPool3dWithIndicesAllNegativeValuesModule_basic",
"MaxPool3dWithIndicesNonDefaultPaddingModule_basic",
Bump Onnx Version to 1.16.1 (#3515) This commit adds the support for new data types: uint4, and int4 and uint8 tensor protos. Also, it moves some tests from failing to crashing. Fixes https://github.com/llvm/torch-mlir/issues/3507 Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-07-02 00:45:45 +08:00
"StdDimEmptyDimModule_basic",
"StdCorrectionEmptyDimModule_basic",
"VarCorrectionEmptyDimModule_basic",
"VarDimEmptyDimModule_basic",
e2e: Enable generate-runtime-verification pass (#3615) This adds the `generate-runtime-verification` pass into the linalg refbackend, and moves all tests that now abort at runtime into the crash set, sorted by their respective errors. I have fixed on set of errors found that way, which are mismatches between the static dimensions we cast to and the actual dynamic dimensions. This was caused by wrong annotations on the test cases, like in https://github.com/llvm/torch-mlir/pull/3615/files#diff-48bfbf41fcad5fa01b49197d251114f84a2b8de4f1d87ab938a061aedd1419b1R1931
2024-08-12 20:15:12 +08:00
# Runtime op verification: rank mismatch in memref.cast
"ViewSizeFromOtherTensor_basic",
[Torch Dialect] support aten.fake_quantize_per_tensor_affine (#3014)
2024-03-15 08:53:29 +08:00
}
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
FX_IMPORTER_TOSA_XFAIL_SET = {
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
"AtenIntMM_basic",
"AtenKthvalueDynamicDimsModule_basic",
"AtenKthvalueFloat64DynamicDimsModule_basic",
"AtenKthvalueFloat64Module_basic",
"AtenKthvalueKeepDimModule_basic",
"AtenKthvalueModule_basic",
"AvgPool3dStaticModule_basic",
"Conv_Transpose1dModule_basic",
"Conv_Transpose1dStaticModule_basic",
"Conv_Transpose2dStaticModule_basic",
"Conv_Transpose3dModule_basic",
"Conv_Transpose3dStaticModule_basic",
"EinsumStaticDiagonalDimensionModule_basic",
"ElementwiseFloatTensorGtIntTensorModule_basic",
"ElementwiseIntTensorLtFloatTensorModule_basic",
"ElementwiseRemainderScalarModule_Bool_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Float_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Float_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDivisor_basic",
"ElementwiseRreluEvalModule_basic",
"ElementwiseRreluEvalStaticModule_basic",
"ElementwiseRreluTrainModule_basic",
"ElementwiseRreluTrainStaticModule_basic",
"FakeQuantizePerTensorAffineCachemaskModule_basic",
"IndexPutWithNoneAndBroadcastModule_basic",
"MaskedScatterStaticBasic_basic",
"MaxUnpool3dModulePad0_basic",
"MaxUnpool3dModule_basic",
"MultinomialModule2D_F32",
"MultinomialModule2D_basic",
"MultinomialModule_basic",
"ReduceAminSingleDim_basic",
"ReduceAminmaxAllDims_basic",
"ReduceAminmaxSingleDim_basic",
"ReduceAnyDimFloatModule_basic",
"RenormModuleFloat16_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
# REMOVE WHEN ENABLE_GQA IS ADDED
"ScaledDotProductAttentionBoolMaskModule_basic",
"ScaledDotProductAttentionDifferentCausalModule_basic",
"ScaledDotProductAttentionSameCausalModule_basic",
"ScaledDotProductAttentionSameDynamicModule_basic",
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
"ScatterAddStaticModule_basic",
"TensorsConcatComplex128FloatModule_basic",
"TensorsConcatComplex128IntModule_basic",
"TensorsConcatComplex64FloatModule_basic",
"TimeOutModule_basic",
"TrilIndicesAllZerosModule_basic",
"TrilIndicesModule_basic",
"TrilIndicesNegativeOffsetModule_basic",
"TrilIndicesOfssetGreaterThanRowModule_basic",
"TriuIndicesAllZerosModule_basic",
"TriuIndicesModule_basic",
"TriuIndicesNegativeOffsetModule_basic",
"TypeConversionUint8ToF32Module_basic",
"WeightNormInterfaceModule_basic",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"AdaptiveAvgPool1dGeneralDynamicNoBatches_basic",
"AdaptiveAvgPool1dGeneralDynamic_basic",
"AdaptiveAvgPool1dStaticLargerOutput_basic",
"AdaptiveAvgPool2dDynamicNoBatch_basic",
"AdaptiveAvgPool2dDynamic_basic",
"AdaptiveAvgPool3dDynamicNoBatch_basic",
"AdaptiveAvgPool3dDynamic_basic",
"AdaptiveMaxPool1dDynamicNoBatch_basic",
"AdaptiveMaxPool1dDynamic_basic",
"AdaptiveMaxPool1dStatic_basic",
"AdaptiveMaxPool2dDynamicNoBatch_basic",
"AdaptiveMaxPool2dDynamicWithIndices_basic",
"AdaptiveMaxPool2dDynamic_basic",
"AdaptiveMaxPool2dStaticWithIndices_basic",
"AdaptiveMaxPool2dStatic_basic",
"AdaptiveMaxPool3dDynamicNoBatch_basic",
"AdaptiveMaxPool3dDynamicWithIndices_basic",
"AdaptiveMaxPool3dDynamic_basic",
"AdaptiveMaxPool3dStaticWithIndices_basic",
"AdaptiveMaxPool3dStatic_basic",
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
"ArangeStartOutViewModule_basic",
"ArgminIntModule_basic",
"ArgminIntModule_multiple_mins",
"ArgminModule_basic",
"ArgminModule_keepDim",
"ArgminModule_with_dim",
"AtenComplexImagModule_basic",
"AtenComplexRealModule_basic",
"AtenComplexViewModule_basic",
"AtenDiagEmbedDefaultDiag_basic",
"AtenDiagEmbedDimDiag_basic",
"AtenDiagEmbedNegOffsetDiag_basic",
"AtenDiagEmbedNonDefault4DDiag_basic",
"AtenDiagEmbedOffsetDiag_basic",
"AtenDiagEmbedRevDimDiag_basic",
"AtenEmbeddingBagStaticModule_basic",
"AtenEmbeddingBagSumExample_basic",
"AtenEyeMModuleInt2D_basic",
"AtenEyeModuleInt2D_basic",
"AtenFloatScalarModule_basic",
Implement lowering for torch.aten.hann_window.periodic (#3502)
2024-07-17 20:51:23 +08:00
"AtenHannWindowPeriodicTrueModule_basic",
"AtenHannWindowPeriodicFalseModule_basic",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
"AtenIntTensorByteDtypeModule_basic",
"AtenIntTensorCharDtypeModule_basic",
"AtenItemFpOpModule_basic",
"AtenItemIntOpModule_basic",
"AtenLinalgCrossBroadcast_basic",
"AtenLinalgCrossCustomDim_basic",
"AtenLinalgCrossDynamic_basic",
"AtenLinalgCrossFloat_basic",
"AtenLinalgCrossInt_basic",
"AtenLinalgCrossNegativeDim_basic",
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
"AtenMatmulQint8VM_basic",
"AtenMatmulQint8VV_basic",
"AtenMatmulQint8_basic",
"AtenMmIntTypes_basic",
"AtenMmQMixedSigni8_basic",
"AtenMmQint8_basic",
"AtenMmQuint8_basic",
"AtenRealView128Module_basic",
"AtenRealView64Module_basic",
"AtenRoundFloatHalfToEvenModule_basic",
"AtenRoundFloatModule_basic",
"AtenSubFloatModule_basic",
"AtenTopKModule_basic",
"AtenTopKSmallestModule_basic",
"Aten_CastLongModule_basic",
"Aten_EmbeddingBagExample_basic",
"AvgPool1dFloatModule_basic",
"AvgPool1dIntModule_basic",
"AvgPool1dStaticModule_basic",
"AvgPool2dCeilModeTrueModule_basic",
"AvgPool2dDivisorOverrideModule_basic",
"AvgPool2dFloatModule_basic",
"AvgPool2dIntModule_basic",
"AvgPool2dStaticModule_basic",
"BernoulliFloatModule_basic",
"BernoulliModule_basic",
"BernoulliOnesModule_basic",
"BernoulliPModule_basic",
"BernoulliTensorModule_basic",
"BernoulliZerosModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BmmIntModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"BroadcastDynamicDimModule_basic",
"BroadcastToModule_basic",
"CeilFloatModule_basic",
"CollapseAllDimensionsModule_basic",
"CollapseFullDynamicModule_basic",
"CollapsePartialDynamicModule_basic",
"CollapseRank1DynamicModule_basic",
"CollapseStaticModule_basic",
"ConstantBoolParameterModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"Conv1dModule_basic",
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"Conv2dWithPaddingDilationStrideStaticModule_grouped",
"Conv2dWithPaddingDilationStrideStaticModule_grouped_multiplier",
"Conv3dModule_basic",
"ConvTbcModule_basic",
"ConvTranspose2DQInt8_basic",
"Conv_Transpose2dModule_basic",
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2DStatic_basic",
"ConvolutionBackwardModule2DStrided_basic",
"ConvolutionBackwardModule2D_basic",
"ConvolutionModule2DGroups_basic",
"ConvolutionModule2DTransposeNonUnitOutputPadding_basic",
"ConvolutionModule2DTransposeStridedStatic_basic",
"ConvolutionModule2DTransposeStrided_basic",
"ConvolutionModule2DTranspose_basic",
"CopyWithDifferentDTypesModule_basic",
"CosineSimilarityStaticBroadcastModule_basic",
"CrossEntropyLossModule_basic",
"CumsumInputDtypeInt32Module_basic",
"CumsumModule_basic",
"CumsumStaticModule_basic",
"CumsumStaticNegativeDimModule_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
"DeformConv2D_basic",
[TorchToLinalg][ONNX] Add Basic Determinant Support (#3481) This adds support for a few ops: - torch.linalg_det - torch._linalg_det (if the LU and pivot returns are unused) - onnx.Det An scf loop is used, since the row reduction algorithm applied here has some loop-carried dependencies. The current support being added here is very basic, and only works if no permutations are required during row reduction, and assumes the matrices are non-singular.
2024-06-26 02:34:19 +08:00
"DeterminantBatchedModule_F32",
"DeterminantDynamicModule_F32",
"DeterminantModule_F32",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"DiagonalModule_basic",
"DiagonalModule_nonsquare",
"DiagonalModule_transposed",
"DiagonalModule_with_dims",
"DiagonalModule_with_dims_and_offset",
"DiagonalModule_with_negative_dims",
"DiagonalModule_with_offset",
"DiagonalWithStaticShapeModule_basic",
"DivFloatModule_basic",
"DivIntModule_basic",
"DropoutTrainModule_basic",
"DropoutTrainStaticShapeModule_basic",
"ElementwiseAcosIntModule_basic",
"ElementwiseAcosModule_basic",
"ElementwiseAcoshIntModule_basic",
"ElementwiseAcoshModule_basic",
"ElementwiseAddScalar_NumToTensorFloat_Module_basic",
"ElementwiseAndScalarModule_basic",
"ElementwiseAndScalarStaticShapeModule_basic",
"ElementwiseAsinIntModule_basic",
"ElementwiseAsinModule_basic",
"ElementwiseAsinhIntModule_basic",
"ElementwiseAsinhModule_basic",
"ElementwiseAtan2FloatIntModule_basic",
"ElementwiseAtan2FloatIntStaticModule_basic",
"ElementwiseAtan2TensorFloatModule_basic",
"ElementwiseAtan2TensorFloatStaticModule_basic",
"ElementwiseAtan2TensorIntModule_basic",
"ElementwiseAtan2TensorIntStaticModule_basic",
"ElementwiseAtanTensorFloatModule_basic",
"ElementwiseAtanTensorIntModule_basic",
"ElementwiseAtanhIntModule_basic",
"ElementwiseAtanhModule_basic",
"ElementwiseAtenFloorDivideBroadcastModule_basic",
"ElementwiseAtenFloorDivideScalarModule_basic",
"ElementwiseAtenFloorDivideScalarNegativeModule_basic",
"ElementwiseAtenFloorDivideTensorNegativeModule_basic",
"ElementwiseAtenFloorDivideTensorPositiveModule_basic",
"ElementwiseAtenLogicalAndOpModule_basic",
"ElementwiseAtenLogicalAndOpPromoteBroadcastModule_basic",
"ElementwiseAtenLogicalAndOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseAtenLogicalNotOpModule_basic",
"ElementwiseAtenLogicalNotOpPromoteModule_basic",
"ElementwiseAtenLogicalXorOpModule_basic",
"ElementwiseAtenLogicalXorOpPromoteBroadcastModule_basic",
"ElementwiseAtenLogicalXorOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseBitwiseAndScalarInt32Module_basic",
"ElementwiseBitwiseAndScalarInt64Module_basic",
"ElementwiseBitwiseAndScalarInt8Module_basic",
"ElementwiseBitwiseLeftShiftInt32Module_basic",
"ElementwiseBitwiseLeftShiftInt64Module_basic",
"ElementwiseBitwiseLeftShiftInt8Module_basic",
"ElementwiseBitwiseRightShiftInt32Module_basic",
"ElementwiseBitwiseRightShiftInt64Module_basic",
"ElementwiseBitwiseRightShiftInt8Module_basic",
"ElementwiseClampMinTensorFloatModule_basic",
"ElementwiseClampMinTensorIntModule_basic",
"ElementwiseClampTensorFloatModule_basic",
"ElementwiseClampTensorIntModule_basic",
"ElementwiseCosIntModule_basic",
"ElementwiseCosModule_basic",
"ElementwiseCoshIntModule_basic",
"ElementwiseCoshModule_basic",
"ElementwiseDequantizePerChannelModule_basic",
"ElementwiseDequantizePerTensorModule_basic",
"ElementwiseDivScalarRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeFloorModule_basic",
"ElementwiseDivScalarRoundingModeFloorStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncModule_basic",
"ElementwiseDivScalarRoundingModeTruncStaticModule_basic",
"ElementwiseDivTensorRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeFloorModule_basic",
"ElementwiseDivTensorRoundingModeFloorStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncModule_basic",
"ElementwiseDivTensorRoundingModeTruncStaticModule_basic",
"ElementwiseErfIntModule_basic",
"ElementwiseErfModule_basic",
"ElementwiseExpIntModule_basic",
"ElementwiseExpm1IntModule_basic",
"ElementwiseExpm1Module_basic",
"ElementwiseGeFloatTensorModule_basic",
"ElementwiseGeIntTensorModule_basic",
"ElementwiseGeluApproximateTanhModule_basic",
"ElementwiseHardshrinkModule_basic",
"ElementwiseHardshrinkStaticModule_basic",
"ElementwiseIntTensorLtFloatScalarModule_basic",
"ElementwiseLeFloatIntScalarModule_basic",
"ElementwiseLeFloatScalarModule_basic",
"ElementwiseLeFloatTensorNanModule_basic",
"ElementwiseLeIntScalarModule_basic",
"ElementwiseLeMixedIntScalarModule_basic",
"ElementwiseLog10IntModule_basic",
"ElementwiseLog10Module_basic",
"ElementwiseLog1pModule_basic",
"ElementwiseLog2IntModule_basic",
"ElementwiseLogIntModule_basic",
"ElementwiseLogSigmoidModule_basic",
"ElementwiseLogitModule_basic",
"ElementwiseMishModule_basic",
"ElementwiseMulTensorComplexDiffModule_basic",
"ElementwiseMulTensorComplexModule_basic",
"ElementwisePowScalarModule_basic",
"ElementwisePowTensorBroadcastModule_basic",
"ElementwisePowTensorBroadcastStaticModule_basic",
"ElementwisePowTensorModule_basic",
"ElementwisePowTensorStaticModule_basic",
"ElementwiseQuantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorUIntModule_basic",
"ElementwiseReciprocalIntModule_basic",
"ElementwiseRemainderScalarModule_Bool_basic",
"ElementwiseRemainderTensorModule_Float_basic",
"ElementwiseRemainderTensorModule_Int_Float_basic",
"ElementwiseRemainderTensorModule_Int_basic",
"ElementwiseRsqrtIntModule_basic",
"ElementwiseSigmoidIntModule_basic",
"ElementwiseSinIntModule_basic",
"ElementwiseSinModule_basic",
"ElementwiseSinhIntModule_basic",
"ElementwiseSinhModule_basic",
"ElementwiseTanIntModule_basic",
"ElementwiseTanModule_basic",
"ElementwiseToDtypeF32ToI64Module_basic",
"ElementwiseToDtypeI64ToUI8Module_basic",
"ElementwiseUnaryIntModule_basic",
"ElementwiseWhereScalarOtherStaticModule_basic",
"EmptyLikeMemoryFormatModule_basic",
"EmptyLikeModule_defaultDtype",
"EmptyLikeModule_falsePinMemory",
"EmptyLikeModule_float",
"EmptyLikeModule_int",
"EmptyModule_contiguous",
"EmptyModule_defaultDtype",
"EmptyModule_falsePinMemory",
"EmptyModule_float",
"EmptyModule_int",
"EmptyModule_uint8",
"EmptyStridedModule_basic",
"EmptyStridedSizeIntStrideModule_basic",
"EqIntModule_basic",
"ExpandModule_basic",
"ExponentialModule_basic",
"FakeQuantizePerTensorAffineDynamicShapeModule_basic",
"FakeQuantizePerTensorAffineModule_basic",
"FakeQuantizePerTensorAffineRoundToEvenModule_basic",
"Fill_TensorFloat32WithFloat32_basic",
"Fill_TensorFloat32WithFloat64_basic",
"Fill_TensorFloat32WithInt64_basic",
"Fill_TensorFloat64WithFloat32Static_basic",
"Fill_TensorFloat64WithFloat32_basic",
"Fill_TensorFloat64WithFloat64_basic",
"Fill_TensorFloat64WithInt64Static_basic",
"Fill_TensorFloat64WithInt64_basic",
"FlipModuleStaticShape_basic",
"FlipModule_basic",
"FlipNegativeIndexModule_basic",
"FloatImplicitModule_basic",
"FullLikeModuleInt2D_basic",
"FullLikeModuleInt3D_basic",
"FullModuleDefaultDtype_basic",
"FullModuleFalsePinMemory_basic",
"FullModuleFloat2D_basic",
"FullModuleFloat3D_basic",
"FullModuleInt2D_basic",
"FullModuleInt3D_basic",
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
"GridSamplerBasic1_basic",
"GridSamplerBasic2_basic",
"GridSamplerBasic3_basic",
"GridSamplerBasic4_basic",
"GtFloatIntModule_basic",
"GtIntModule_basic",
"HBC_basic",
"IndexPut1DFloatAccumulateModule_basic",
"IndexPut1DFloatNonAccumulateModule_basic",
"IndexPut1DIntAccumulateModule_basic",
"IndexPut1DIntNonAccumulateModule_basic",
"IndexPut2DFloatAccumulateModule_basic",
"IndexPut2DFloatNonAccumulateModule_basic",
"IndexPut2DIntAccumulateModule_basic",
"IndexPut2DIntNonAccumulateModule_basic",
"IndexPut3DFloatAccumulateModule_basic",
"IndexPut3DFloatNonAccumulateModule_basic",
"IndexPut3DIntAccumulateModule_basic",
"IndexPut3DIntNonAccumulateModule_basic",
"IndexPutHackedTwin1DFloatAccumulateModule_basic",
"IndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin1DIntAccumulateModule_basic",
"IndexPutHackedTwin1DIntNonAccumulateModule_basic",
"IndexPutHackedTwin2DFloatAccumulateModule_basic",
"IndexPutHackedTwin2DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin2DIntAccumulateModule_basic",
"IndexPutHackedTwin2DIntNonAccumulateModule_basic",
"IndexPutHackedTwin3DFloatAccumulateModule_basic",
"IndexPutHackedTwin3DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin3DIntAccumulateModule_basic",
"IndexPutHackedTwin3DIntNonAccumulateModule_basic",
"IndexPutImpl1DFloatAccumulateModule_basic",
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
"IndexPutImpl2DFloatAccumulateModule_basic",
"IndexPutImpl2DFloatNonAccumulateModule_basic",
"IndexPutImpl2DImplicitModule_basic",
"IndexPutImpl2DIndexModule_basic",
"IndexPutImpl2DNoneIndexStaticModule_basic",
"IndexPutImpl3DFloatAccumulateModule_basic",
"IndexPutImpl3DFloatNonAccumulateModule_basic",
"IndexPutImplIndexWithNoneModule_basic",
"IndexSelectDynamicIndexSizeModule_basic",
"IndexSelectDynamicInputSizeModule_basic",
"IndexSelectDynamicModulebasic",
"IndexSelectNegativeDimModule_basic",
"IndexSelectRank0IdxModule_basic",
"IndexSelectSingleIdxModule_basic",
"IndexSelectTwoIdxModule_basic",
"IndexSelectWholeDimensionModule_basic",
"IndexSelectWholeTensorModule_basic",
"IndexTensorNegativeIndexModule_basic",
Modifies onnx resize lowering to fix numerical issues (#3381) Updates: - some unsupported modes are now going to report a match failure for unsupported coordinate transformation modes. - fixes a bug that was introduced in the last patch for resize (my bad...) - uses actual x and y coordinates for computing weights in bilinear interpolation (rather than eps modified values) - slightly simplifies the bilinear interpolation payload for readability and performance - passes coordinate transformation mode information from an onnx.Resize op to the mode string for the aten._interpolate op. This allows us to perform custom logic in the torch->linalg lowering to support onnx.Resize options without losing the default behaviors of the interpolate op.
2024-05-31 08:34:37 +08:00
"InterpolateDynamicModule_sizes_bilinear",
"InterpolateDynamicModule_sizes_nearest",
"InterpolateStaticModule_scales_bilinear_align_corners",
"InterpolateDynamicModule_scales_recompute_bilinear",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"IntFloatModule_basic",
"IntImplicitModule_basic",
"IsFloatingPointFloat_True",
"IsFloatingPointInt_False",
"LenStrModule_basic",
"LinalgNormKeepDimComplexModule_basic",
"LinalgVectorNormComplexModule_basic",
"LinspaceDtypeModule_basic",
"LinspaceEmptyModule_basic",
"LinspaceModule_basic",
"LinspaceOneSizeModule_basic",
"LinspaceTwoSizeModule_basic",
"MaskedFillTensorFloatValueModule_basic",
"MatmulBroadcastBatchDim_basic",
"MatmulStaticBroadcast_basic",
"MaxPool1dCeilModeTrueModule_basic",
"MaxPool1dModule_basic",
"MaxPool2dCeilModeTrueModule_basic",
"MaxPool2dModule_basic",
"MaxPool2dWithIndicesAllNegativeValuesModule_basic",
"MaxPool2dWithIndicesAllOnesModule_basic",
"MaxPool2dWithIndicesBackwardDynamic3DModule_basic",
"MaxPool2dWithIndicesBackwardDynamic4DModule_basic",
"MaxPool2dWithIndicesBackwardStatic3DModule_basic",
"MaxPool2dWithIndicesBackwardStatic4DModule_basic",
"MaxPool2dWithIndicesCeilModeTrueModule_basic",
"MaxPool2dWithIndicesFullSizeKernelModule_basic",
"MaxPool2dWithIndicesModule_basic",
"MaxPool2dWithIndicesNonDefaultDilationModule_basic",
"MaxPool2dWithIndicesNonDefaultPaddingModule_basic",
"MaxPool2dWithIndicesNonDefaultParamsModule_basic",
"MaxPool2dWithIndicesNonDefaultStrideModule_basic",
"MaxPool2dWithIndicesStaticModule_basic",
"MaxPool3dCeilModeTrueModule_basic",
"MaxPool3dEmptyStrideStaticModule_basic",
"MaxPool3dLargeDatadModule_basic",
"MaxPool3dModuleRandomSimple_basic",
"MaxPool3dModule_basic",
"MaxPool3dStaticCeilModeTrueModule_basic",
"MaxPool3dStaticModule_basic",
[TorchToLinalg] Support lowering MaxPool3dWithIndices (#3652) Support torch.MaxPool3dWithIndices lowering to linalg backend.
2024-08-28 03:14:25 +08:00
"MaxPool3dWithIndicesAllNegativeValuesModule_basic",
"MaxPool3dWithIndicesAllOnesModule_basic",
"MaxPool3dWithIndicesCeilModeTrueModule_basic",
"MaxPool3dWithIndicesFullSizeKernelModule_basic",
"MaxPool3dWithIndicesModule_basic",
"MaxPool3dWithIndicesNonDefaultDilationModule_basic",
"MaxPool3dWithIndicesNonDefaultPaddingModule_basic",
"MaxPool3dWithIndicesNonDefaultParamsModule_basic",
"MaxPool3dWithIndicesNonDefaultStrideModule_basic",
"MaxPool3dWithIndicesStaticModule_basic",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"MeanDimDtypeModule_basic",
"MeanDimEmptyDimModule_basic",
"MeanDimNoneDimModule_basic",
"MeanDtypeModule_basic",
"MseLossMeanReductionModule_basic",
"MseLossSumReductionWithDifferentElemTypeModule_basic",
"MulFloatModule_basic",
"MulIntModule_basic",
"NativeBatchNorm1DModule_basic",
"NativeBatchNorm2DModule_basic",
"NativeBatchNorm3DModule_basic",
"NativeBatchNormNoneWeightModule_basic",
"NativeDropoutTrainModule_basic",
"NativeDropoutTrainStaticShapeModule_basic",
"NativeGroupNormBackwardModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
"NewEmptyModuleDefaultDtype_basic",
"NewEmptyModuleFalsePinMemory_basic",
"NewEmptyModuleFloat2D_basic",
"NewEmptyModuleFloat3D_basic",
"NewEmptyModuleInt2D_basic",
"NewEmptyModuleInt3D_basic",
"NewEmptyModuleLayoutIntDtype_basic",
"NewEmptyModuleNonDefaultFloatDtype_basic",
"NewEmptyModuleNonDefaultIntDtype_basic",
"NewEmptyStridedModuleDefaultDtype_basic",
"NewFullModuleInt2D_basic",
"NewFullModuleInt3D_basic",
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
"NllLossModuleBackwardMeanWeight_basic",
"NllLossModuleBackwardMean_basic",
"NllLossModuleBackwardSumWeight_basic",
"NllLossModuleBackwardSum_basic",
"NllLossModuleBackwardWeight_basic",
"NllLossModuleBackward_basic",
"NllLossModuleBackward_ignore_index",
"NllLossModule_1D_basic",
"NllLossModule_mean_basic",
"NllLossModule_sum_basic",
"NormScalarComplexModule_basic",
"NormScalarModule_basic",
"NormScalarOptDimKeepDimComplexModule_basic",
"NormalFunctionalModule_basic",
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"OnesLikeModule_falsePinMemory",
"PixelShuffleModuleFullDynamic_basic",
"PixelShuffleModuleSpatiallyDynamic_basic",
"PixelShuffleModuleSpatiallyStatic_basic",
"PixelShuffleModuleStaticRank3Int64_basic",
"PixelShuffleModuleStaticRank4Float32_basic",
"PowIntFloatModule_basic",
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsConvertElementTypeModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
"QuantizedBatchedInputSingleLayer_basic",
"QuantizedMLP_basic",
"QuantizedNoLayer_basic",
"QuantizedReluInt32_basic",
"QuantizedReluInt8_basic",
"QuantizedReluUint8_basic",
"QuantizedSingleLayer_basic",
"RandIntDtypeModule_basic",
"RandIntLowDtypeModule_basic",
"RandIntLowModule_basic",
"RandIntModule_basic",
"RandIntPinMemoryModule_basic",
"RandLikeDtypeModule_basic",
"RandLikeModule_basic",
"RandModule_basic",
"RandnDtypeDeviceModule_basic",
"RandnGeneratorF64Module_basic",
"RandnGeneratorModule_basic",
"RandnLikeDtypeModule_basic",
"RandnLikeModule_basic",
"RandnModule_basic",
"ReduceAllDimBool_basic",
"ReduceAllDimEmpty_basic",
"ReduceAllDimFloat_basic",
"ReduceAllDimInt_basic",
"ReduceAllFloatModule_basic",
"ReduceAllIntModule_basic",
"ReduceAnyFloatModule_basic",
"ReduceAnyIntModule_basic",
"ReduceFrobeniusNormComplexModule_basic",
"ReduceL1NormComplexModule_basic",
"ReduceL1NormWithDTypeModule_basic",
"ReduceL2NormComplexModule_basic",
"ReduceL3NormAllDimsModule_basic",
"ReduceL3NormKeepDimComplexModule_basic",
"ReduceL3NormKeepDimModule_basic",
"ReduceMaxAllDims_basic",
"ReduceMaxAlongDimUnsignedInt_basic",
"ReduceMaxFloatModule_basic",
"ReduceMaxSignedIntModule_basic",
"ReduceMaxUnsignedIntModule_basic",
"ReduceMinAlongDimNegative_basic",
"ReduceMinAlongDimSignedInt_basic",
"ReduceMinAlongDimUnsignedInt_basic",
"ReduceMinAlongDim_basic",
"ReduceMinFloatModule_basic",
"ReduceMinKeepDimReturnBoth_basic",
"ReduceMinKeepDim_basic",
"ReduceMinSignedIntModule_basic",
"ReduceMinUnsignedIntModule_basic",
"ReduceProdDimIntFloatModule_basic",
"ReduceProdDtypeFloatModule_basic",
"ReduceProdDtypeIntModule_basic",
"ReduceProdElementTypeBoolModule_basic",
"ReduceProdFloatModule_basic",
"ReduceProdSignedIntModule_basic",
"ReduceProdUnsignedIntModule_basic",
"ReduceSumDimIntListDtypeFloatModule_basic",
"ReduceSumDimIntListDtypeIntModule_basic",
"ReduceSumDimIntListElementTypeBoolModule_basic",
"ReduceSumDimIntListEmptyDimModule_basic",
"ReduceSumDtypeFloatModule_basic",
"ReduceSumDtypeIntModule_basic",
"ReduceSumElementTypeBoolModule_basic",
"ReflectionPad1dModule2dInput_Right",
"ReflectionPad1dModule2dInput_basic",
"ReflectionPad1dModule3dInput_Left",
"ReflectionPad1dModule3dInput_basic",
"ReflectionPad2dModule_Bottom",
"ReflectionPad2dModule_Left",
"ReflectionPad2dModule_Right",
"ReflectionPad2dModule_Top",
"ReflectionPad2dModule_basic",
"ReplicationPad2dModule_basic",
"ReplicationPad2dModule_bottom0",
"ReplicationPad2dModule_left0",
"ReplicationPad2dModule_right0",
"ReplicationPad2dModule_top0",
"RollModule_basic",
"RsubInt0d_NumToTensor_Module_basic",
"RsubIntModule_basic",
"RsubIntModule_noalpha_basic",
"ScalarConstantTupleModule_basic",
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
"ScatterReduceFloatMaxModule",
"ScatterReduceFloatMaxModuleIncludeSelf",
"ScatterReduceFloatMeanModule",
"ScatterReduceFloatMeanModuleIncludeSelf",
"ScatterReduceFloatMinModule",
"ScatterReduceFloatMinModuleIncludeSelf",
"ScatterReduceFloatProdModule",
"ScatterReduceFloatProdModuleIncludeSelf",
"ScatterReduceFloatSumModule",
"ScatterReduceFloatSumModuleIncludeSelf",
"ScatterReduceIntMaxModule",
"ScatterReduceIntMaxModuleIncludeSelf",
"ScatterReduceIntMeanModule",
"ScatterReduceIntMeanModuleIncludeSelf",
"ScatterReduceIntMinModule",
"ScatterReduceIntMinModuleIncludeSelf",
"ScatterReduceIntProdModule",
"ScatterReduceIntProdModuleIncludeSelf",
"ScatterReduceIntSumModule",
"ScatterReduceIntSumModuleIncludeSelf",
"ScatterSrcModule_basic",
"ScatterSrcStaticModule_basic",
"ScatterValueFloatModule_basic",
"ScatterValueIntModule_basic",
"SelectScattertModule_basic",
"SelectScattertStaticModule_basic",
implement lowering of torch.aten._linalg_slogdet (#3524)
2024-07-19 13:54:43 +08:00
"SignAndLogarithmOfDeterminantModule_F32",
"SignAndLogarithmOfDeterminantBatchedModule_F32",
"SignAndLogarithmOfDeterminantDynamicModule_F32",
[FxImporter][TOSA] Enable FxImporter to TOSA e2e tests (#3349) Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-16 05:37:30 +08:00
"SliceCopyEndGreaterThanDimSize_Module_basic",
"SliceCopyNegative_Module_basic",
"SliceCopyNonZeroDim_Module_basic",
"SliceCopyStartGreaterThanDimSize_Module_basic",
"SliceCopy_Module_basic",
"SliceEndSleStartModule_basic",
"SliceOutOfLowerBoundEndIndexModule_basic",
"SliceOutOfLowerBoundStartIndexModule_basic",
"SliceScatterModule_basic",
"SliceScatterNegativeDimModule_basic",
"SliceScatterNegativeEndModule_basic",
"SliceScatterStaticModule_basic",
"SliceScatterStepVariationModule_basic",
"SliceScatterZeroDimModule_basic",
"SliceSizeTwoStepModule_basic",
"SoftplusModule_basic",
"SortIntListReverse_basic",
"SortIntList_basic",
"SortTensorDescending_basic",
"SortTensorInteger_basic",
"SortTensorNegativeDimension_basic",
"SortTensorSpecificDimension_basic",
"SortTensor_basic",
"SplitDimDynamicModule_basic",
"SplitDimStaticModule_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SubFloatModule_basic",
"SubIntModule_basic",
"TModuleRank0_basic",
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"TensorToIntZeroRank_basic",
"TensorToInt_basic",
"TensorsConcatPromoteDTypeModule_basic",
"TensorsStackPromoteDTypeModule_basic",
"TestMultipleTensorAndPrimitiveTypesReturn_basic",
"Threshold1dIntModule_basic",
"Threshold2dIntModule_basic",
"Threshold3dIntModule_basic",
"ThresholdBackward1dFloatModule_basic",
"ThresholdBackward1dIntModule_basic",
"ThresholdBackward1dMixedModule_basic",
"ThresholdBackward2dFloatModule_basic",
"ThresholdBackward2dIntModule_basic",
"ThresholdBackward2dMixedModule_basic",
"ThresholdBackward3dFloatModule_basic",
"ThresholdBackward3dIntModule_basic",
"ThresholdBackward3dMixedModule_basic",
"ToCopyWithDTypeFalsePinMemoryModule_basic",
"ToCopyWithDTypeModule_basic",
"TorchPrimLoopForLikeModule_basic",
"TorchPrimLoopWhileLikeModule_basic",
"TraceModule_basic",
"TraceModule_empty",
"TraceModule_nonsquare",
"TraceSignedIntModule_basic",
"TraceUnsignedIntModule_basic",
"TraceUnsignedIntModule_empty",
"TypeConversionI1ToF64Module_basic",
"TypeConversionI1ToI32Module_basic",
"UniformModule_basic",
"UniformNoCorrelationModule_basic",
"UniformStaticShapeModule_basic",
"UnsafeIndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"UpSampleNearest2dBackwardScalesNone_basic",
"UpSampleNearest2dBackward_basic",
"UpSampleNearest2dDynamicFactor_basic",
"UpSampleNearest2dDynamicSize_basic",
"UpSampleNearest2dStaticFactor_basic",
"UpSampleNearest2dStaticSize_basic",
"UpSampleNearest2d_basic",
"VarMeanBiasedModule_basic",
"VarMeanCorrectionNoneModule_basic",
"VarMeanUnbiasedModule_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewSizeFromOtherTensor_basic",
"ZeroFloat32Module_basic",
"ZeroInt32Module_basic",
"ZeroInt64Module_basic",
"ZerosLikeModule_falsePinMemory",
}
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
ONNX_TOSA_CRASHING_SET = {
"StdCorrectionEmptyDimModule_basic",
"StdDimEmptyDimModule_basic",
"VarCorrectionEmptyDimModule_basic",
"VarDimEmptyDimModule_basic",
"ViewSizeFromOtherTensor_basic",
}
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
ONNX_TOSA_XFAIL_SET = {
[TOSA] Add Torch to Tosa Legalization for torch.tril (#3678) Change-Id: Ie5ba31a27394c3adcea00266a9d562862dbd8b08 Signed-off-by: Justin Ngo <justin.ngo@arm.com>
2024-09-06 02:27:29 +08:00
"AdaptiveAvgPool2dFixedKernelStrideSizeStaticModule_basic",
"AdaptiveAvgPool2dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool2dOutputSizeDivisibleByInputStaticModule_basic",
"ArgmaxKeepdimModule_basic",
"AtenIntMM_basic",
"AtenKthvalueDynamicDimsModule_basic",
"AtenKthvalueFloat64DynamicDimsModule_basic",
"AtenKthvalueFloat64Module_basic",
"AtenKthvalueKeepDimModule_basic",
"AtenKthvalueModule_basic",
"AvgPool2dCountIncludePadFalseStaticModule_basic",
"AvgPool3dStaticModule_basic",
"Conv_Transpose1dModule_basic",
"Conv_Transpose1dStaticModule_basic",
"Conv_Transpose2dStaticModule_basic",
"Conv_Transpose3dModule_basic",
"Conv_Transpose3dStaticModule_basic",
"EinsumStaticDiagonalDimensionModule_basic",
"EinsumStaticModule_basic",
"ElementwiseFmaxModule_basic",
"ElementwiseFminModule_basic",
"ElementwiseGeluApproximateTanhModule_basic",
"ElementwiseIntTensorLtFloatTensorModule_basic",
"ElementwiseNanToNumWithNoneModule_Basic",
"ElementwiseRad2DegIntModule_basic",
"ElementwiseRad2DegModule_basic",
"ElementwiseRemainderScalarModule_Bool_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_Float_NegativeDivisor_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDividend_basic",
"ElementwiseRemainderScalarModule_Int_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_Float_NegativeDivisor_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDividend_basic",
"ElementwiseRemainderTensorModule_Int_NegativeDivisor_basic",
"ElementwiseRreluTrainModule_basic",
"ElementwiseRreluTrainStaticModule_basic",
"FakeQuantizePerTensorAffineCachemaskModule_basic",
"IndexPutWithNoneAndBroadcastModule_basic",
"MaskedScatterStaticBasic_basic",
"MaxUnpool3dModulePad0_basic",
"MaxUnpool3dModule_basic",
"MultinomialModule2D_F32",
"MultinomialModule2D_basic",
"MultinomialModule_basic",
"ReduceAmaxEmptyDim_basic",
"ReduceAminSingleDim_basic",
"ReduceAminmaxAllDims_basic",
"ReduceAminmaxSingleDim_basic",
"ReduceAnyDimFloatModule_basic",
"RenormModuleFloat16_basic",
"RenormModuleFloat32DynamicDims_basic",
"RenormModuleFloat32NegativeDim_basic",
"RenormModuleFloat32_basic",
"ScatterAddStaticModule_basic",
"TensorSplitSections_GetItemModule_basic",
"TensorSplitSections_ListUnpackModule_basic",
"TensorsConcatComplex128FloatModule_basic",
"TensorsConcatComplex128IntModule_basic",
"TensorsConcatComplex64FloatModule_basic",
"TimeOutModule_basic",
"TypeConversionUint8ToF32Module_basic",
"UnfoldModule_basic",
"WeightNormInterfaceModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"AdaptiveAvgPool1dGeneralDynamicNoBatches_basic",
"AdaptiveAvgPool1dGeneralDynamic_basic",
"AdaptiveAvgPool1dNonUnitOutputSizeDynamicModule_basic",
"AdaptiveAvgPool1dNonUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool1dStaticEvenMultiple_basic",
"AdaptiveAvgPool1dStaticLargerOutput_basic",
"AdaptiveAvgPool1dUnitOutputSizeDynamicModule_basic",
"AdaptiveAvgPool1dUnitOutputSizeStaticModule_basic",
"AdaptiveAvgPool2dDynamicNoBatch_basic",
"AdaptiveAvgPool2dDynamic_basic",
"AdaptiveAvgPool2dNonUnitOutputSizeDynamicModule_basic",
"AdaptiveAvgPool2dOutputSizeDivisibleByInputDynamicModule_basic",
"AdaptiveAvgPool2dUnitOutputSizeDynamicModule_basic",
"AdaptiveAvgPool3dDynamicNoBatch_basic",
"AdaptiveAvgPool3dDynamic_basic",
"AdaptiveMaxPool1dDynamicNoBatch_basic",
"AdaptiveMaxPool1dDynamic_basic",
"AdaptiveMaxPool1dStatic_basic",
"AdaptiveMaxPool2dDynamicNoBatch_basic",
"AdaptiveMaxPool2dDynamicWithIndices_basic",
"AdaptiveMaxPool2dDynamic_basic",
"AdaptiveMaxPool2dStaticWithIndices_basic",
"AdaptiveMaxPool2dStatic_basic",
"AdaptiveMaxPool3dDynamicNoBatch_basic",
"AdaptiveMaxPool3dDynamicWithIndices_basic",
"AdaptiveMaxPool3dDynamic_basic",
"AdaptiveMaxPool3dStaticWithIndices_basic",
"AdaptiveMaxPool3dStatic_basic",
"AddCDivModule_basic",
"AddIntModule_basic",
[TorchToArith] Add a lowering for `torch.add.float_int` (#3594)
2024-08-08 00:55:27 +08:00
"AddFloatIntModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"AddSizeIntModule_basic",
"AddSizeIntNegDimModule_basic",
"Add_MixPModule_basic",
"Add_Module_basic",
"AddmmModuleFloat_basic",
"AddmmModule_broadcastable",
"AddmmModule_differentRankBroadcastable",
"AllBoolFalseModule_basic",
"AllBoolTrueModule_basic",
"AnyBoolFalseModule_basic",
"AnyBoolTrueModule_basic",
"ArangeStartOutDtypeModule_basic",
"ArangeStartOutViewModule_basic",
"ArgmaxIntModule_basic",
"ArgmaxIntModule_multiple_maxs",
"ArgmaxModule_basic",
"ArgmaxModule_with_dim",
"ArgminIntModule_basic",
"ArgminIntModule_multiple_mins",
"ArgminModule_basic",
"ArgminModule_keepDim",
"ArgminModule_with_dim",
"AtenComplex64Module_basic",
"AtenComplexImagModule_basic",
"AtenComplexRealModule_basic",
"AtenComplexViewModule_basic",
"AtenDiagEmbedDefaultDiag_basic",
"AtenDiagEmbedDimDiag_basic",
"AtenDiagEmbedNegOffsetDiag_basic",
"AtenDiagEmbedNonDefault4DDiag_basic",
"AtenDiagEmbedOffsetDiag_basic",
"AtenDiagEmbedRevDimDiag_basic",
"AtenEmbeddingBagStaticModule_basic",
"AtenEmbeddingBagSumExample_basic",
"AtenFloatScalarModule_basic",
"AtenIntBoolOpConstFalseModule_basic",
"AtenIntBoolOpConstTrueModule_basic",
"AtenIntBoolOpModule_basic",
"AtenIntTensorByteDtypeModule_basic",
"AtenIntTensorCharDtypeModule_basic",
"AtenItemFpOpModule_basic",
"AtenItemIntOpModule_basic",
"AtenLinalgCrossDynamic_basic",
"AtenMatmulQMixedSigni8Transpose_basic",
"AtenMatmulQMixedSigni8_basic",
"AtenMatmulQint8MV_basic",
"AtenMatmulQint8VM_basic",
"AtenMatmulQint8VV_basic",
"AtenMatmulQint8_basic",
"AtenMmFloatTypes_basic",
"AtenMmIntTypes_basic",
"AtenMmQMixedSigni8_basic",
"AtenMmQint8_basic",
"AtenMmQuint8_basic",
[MLIR][TORCH] Add E2E support for aten.polar op (#3671) Signed-Off By: Vivek Khandelwal <vivekkhandelwal1424@gmail.com>
2024-09-03 13:21:03 +08:00
"AtenPolarFloatModule_basic",
"AtenPolarDoubleModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"AtenRealView128Module_basic",
"AtenRealView64Module_basic",
"AtenRoundFloatHalfToEvenModule_basic",
"AtenRoundFloatModule_basic",
"AtenSubFloatModule_basic",
"AtenTopKModule_basic",
"AtenTopKSmallestModule_basic",
"AtenTrilModule_basic",
"AtenTrilWithNegDiagonalModule_basic",
"AtenTrilWithPosDiagonalModule_basic",
"AtenTriuModule_basic",
"AtenTriuWithNegDiagonalModule_basic",
"AtenTriuWithPosDiagonalModule_basic",
"Aten_CastLongModule_basic",
"Aten_EmbeddingBagExample_basic",
"AvgPool1dFloatModule_basic",
"AvgPool1dIntModule_basic",
"AvgPool1dStaticModule_basic",
"AvgPool2dCeilModeTrueModule_basic",
"AvgPool2dDivisorOverrideModule_basic",
"AvgPool2dFloatModule_basic",
"AvgPool2dIntModule_basic",
"AvgPool2dStaticModule_basic",
"AvgPool2dWithoutPadModule_basic",
"BatchMlpLayerModule_basic",
"BernoulliFloatModule_basic",
"BernoulliModule_basic",
"BernoulliOnesModule_basic",
"BernoulliPModule_basic",
"BernoulliTensorModule_basic",
"BernoulliZerosModule_basic",
"BincountMinlengthModule_basic",
"BincountModule_basic",
"BincountStaticSizeModule_basic",
"BmmIntModule_basic",
"BoolFloatConstantModule_basic",
"BoolFloatFalseModule_basic",
"BoolFloatTrueModule_basic",
"BoolIntConstantModule_basic",
"BoolIntFalseModule_basic",
"BoolIntTrueModule_basic",
"BoolTensorHandleSignless_basic",
"BroadcastDynamicDimModule_basic",
"BroadcastToModule_basic",
"BucketizeTensorFloatModule_basic",
"BucketizeTensorModule_basic",
"BucketizeTensorOutInt32RightModule_basic",
"BucketizeTensorStaticFloatModule_basic",
"BucketizeTensorStaticModule_basic",
"CeilFloatModule_basic",
"ChunkListUnpackDynamic_Module_basic",
"ChunkListUnpackUnevenDynamic_Module_basic",
"ChunkListUnpackUneven_Module_basic",
"ChunkListUnpack_Module_basic",
"CollapseAllDimensionsModule_basic",
"CollapseFullDynamicModule_basic",
"CollapsePartialDynamicModule_basic",
"CollapseRank1DynamicModule_basic",
"CollapseStaticModule_basic",
"ConstantBoolParameterModule_basic",
"ConstantPad2dStaticModule_basic",
"ConstantPadNdModule_basic",
"ConstantPadNdPartialStaticModule_basic",
"ConstantPadNdStaticModule_basic",
"ContainsIntList_False",
"ContainsIntList_True",
"Conv1dModule_basic",
"Conv2dBiasNoPaddingModule_basic",
"Conv2dModule_basic",
"Conv2dNoPaddingModule_basic",
"Conv2dQInt8Module_basic",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_depthwise",
[TorchToLinalg] add support for quantized group conv (#3341) This addresses 7 of the model failures I'm seeing in the test suite. See [Shark-Turbine issue #566](https://github.com/nod-ai/SHARK-Turbine/issues/566). Need the op ```linalg.conv_2d_ngchw_gfchw_q``` to be added upstream before merging this. See [llvm-project PR #92136 ](https://github.com/llvm/llvm-project/pull/92136). A small additional expansion to operand quantization is included in this patch to address a model failure that occurs when unblocking the quantized group convolutions in one of these onnx models.
2024-06-04 00:27:44 +08:00
"Conv2dQInt8Module_grouped",
[TorchToLinalg] add support for depthwise qconv (#3564) - Adds support for lowering depthwise + quantized convolution ops to linalg::DepthwiseConv2DNhwcHwcQOp - Changed the variable name for groupSize (which is really C/G) to the more appropriate numGroups (G). - Discovered in e2e testing that linalg does not accept (Cin = groups && Cout = K*groups for K>1) as a "depthwise" conv, so this also updates the case-checking to reflect this issue.
2024-07-30 03:25:07 +08:00
"Conv2dQInt8Module_not_depthwise",
[torch] Basic support for per-channel quantized graphs (#3623) This patch adds basic support for lowering graphs with per-channel quantization. Per-channel quantized ops have to be excluded from `FuseQuantizedOps` for now but can be used in QDQ quantized form. Using this patch, we're able to import and execute (on the linalg backend) graphs with per-channel quantization applied using the "new" PyTorch 2.0 Export Quantization.
2024-08-10 21:51:09 +08:00
"Conv2dQInt8PerChannelModule_basic",
"Conv2dQInt8PerChannelModule_depthwise",
"Conv2dQInt8PerChannelModule_grouped",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"Conv2dWithPaddingDilationStrideModule_basic",
"Conv2dWithPaddingDilationStrideStaticModule_grouped",
"Conv2dWithPaddingDilationStrideStaticModule_grouped_multiplier",
"Conv2dWithPaddingModule_basic",
"Conv3dModule_basic",
"ConvTbcModule_basic",
"ConvTranspose2DQInt8_basic",
"Conv_Transpose2dModule_basic",
"Convolution2DModule_basic",
"Convolution2DStridedModule_basic",
"ConvolutionBackwardModule2DPadded_basic",
"ConvolutionBackwardModule2DStatic_basic",
"ConvolutionBackwardModule2DStrided_basic",
"ConvolutionBackwardModule2D_basic",
"ConvolutionModule2DGroups_basic",
"ConvolutionModule2DTransposeNonUnitOutputPadding_basic",
"ConvolutionModule2DTransposeStridedStatic_basic",
"ConvolutionModule2DTransposeStrided_basic",
"ConvolutionModule2DTranspose_basic",
"CopyModule_basic",
"CopyWithDifferentDTypesAndSizesModule_basic",
"CopyWithDifferentDTypesModule_basic",
"CopyWithDifferentSizesModule_basic",
"CosineSimilarityStaticBroadcastModule_basic",
"CrossEntropyLossModule_basic",
"CrossEntropyLossNoReductionModule_basic",
"CumsumInputDtypeInt32Module_basic",
"CumsumModule_basic",
"CumsumStaticModule_basic",
"CumsumStaticNegativeDimModule_basic",
[ONNX] Basic Support for DeformConv (#3469) This adds a torchvision op to torch-mlir and a path from onnx.DeformConv to torchvision.deform_conv2d. I'm not implementing the torch->linalg lowering for the torchvision op yet, but posting this PR to get feedback on some of the choices being made here and to flesh out the onnx frontend a bit.
2024-06-26 01:16:51 +08:00
"DeformConv2D_basic",
[TorchToLinalg][ONNX] Add Basic Determinant Support (#3481) This adds support for a few ops: - torch.linalg_det - torch._linalg_det (if the LU and pivot returns are unused) - onnx.Det An scf loop is used, since the row reduction algorithm applied here has some loop-carried dependencies. The current support being added here is very basic, and only works if no permutations are required during row reduction, and assumes the matrices are non-singular.
2024-06-26 02:34:19 +08:00
"DeterminantModule_F32",
"DeterminantBatchedModule_F32",
"DeterminantDynamicModule_F32",
"DeterminantModule_F32",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"DiagonalModule_basic",
"DiagonalModule_nonsquare",
"DiagonalModule_transposed",
"DiagonalModule_with_dims",
"DiagonalModule_with_dims_and_offset",
"DiagonalModule_with_negative_dims",
"DiagonalModule_with_offset",
"DiagonalWithStaticShapeModule_basic",
"DivFloatModule_basic",
"DivIntModule_basic",
"DropoutTrainModule_basic",
"DropoutTrainStaticShapeModule_basic",
"ElementwiseAcosIntModule_basic",
"ElementwiseAcosModule_basic",
"ElementwiseAcoshIntModule_basic",
"ElementwiseAcoshModule_basic",
"ElementwiseAddScalarInt64Module_basic",
"ElementwiseAddScalarIntModule_basic",
"ElementwiseAsinIntModule_basic",
"ElementwiseAsinModule_basic",
"ElementwiseAsinhIntModule_basic",
"ElementwiseAsinhModule_basic",
"ElementwiseAtan2FloatIntModule_basic",
"ElementwiseAtan2FloatIntStaticModule_basic",
"ElementwiseAtan2TensorFloatModule_basic",
"ElementwiseAtan2TensorFloatStaticModule_basic",
"ElementwiseAtan2TensorIntModule_basic",
"ElementwiseAtan2TensorIntStaticModule_basic",
"ElementwiseAtanTensorFloatModule_basic",
"ElementwiseAtanTensorIntModule_basic",
"ElementwiseAtanhIntModule_basic",
"ElementwiseAtanhModule_basic",
"ElementwiseAtenDivIntScalarModule_basic",
"ElementwiseAtenFloorDivideBroadcastModule_basic",
"ElementwiseAtenFloorDivideScalarModule_basic",
"ElementwiseAtenFloorDivideScalarNegativeModule_basic",
"ElementwiseAtenFloorDivideTensorNegativeModule_basic",
"ElementwiseAtenFloorDivideTensorPositiveModule_basic",
"ElementwiseAtenIsneginfOpModule_basic",
"ElementwiseAtenIsposinfOpModule_basic",
"ElementwiseAtenLogicalAndOpModule_basic",
"ElementwiseAtenLogicalAndOpPromoteBroadcastModule_basic",
"ElementwiseAtenLogicalAndOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseAtenLogicalNotOpPromoteModule_basic",
"ElementwiseAtenLogicalOrOpBrodcastModule_basic",
"ElementwiseAtenLogicalOrOpDiffArgs1Module_basic",
"ElementwiseAtenLogicalOrOpDiffArgs2Module_basic",
"ElementwiseAtenLogicalOrOpDiffArgs3Module_basic",
"ElementwiseAtenLogicalOrOpNegativeModule_basic",
"ElementwiseAtenLogicalOrOpRandomFloatModule_basic",
"ElementwiseAtenLogicalOrOpRandomModule_basic",
"ElementwiseAtenLogicalXorOpModule_basic",
"ElementwiseAtenLogicalXorOpPromoteBroadcastModule_basic",
"ElementwiseAtenLogicalXorOpPromoteBroadcastStaticShapeModule_basic",
"ElementwiseBitwiseAndModule_basic",
"ElementwiseBitwiseLeftShiftInt32Module_basic",
"ElementwiseBitwiseLeftShiftInt64Module_basic",
"ElementwiseBitwiseLeftShiftInt8Module_basic",
"ElementwiseBitwiseNotInt32Module_basic",
"ElementwiseBitwiseNotInt64Module_basic",
"ElementwiseBitwiseOrModule_basic",
"ElementwiseBitwiseOrStaticShapeModule_basic",
"ElementwiseBitwiseRightShiftInt32Module_basic",
"ElementwiseBitwiseRightShiftInt64Module_basic",
"ElementwiseBitwiseRightShiftInt8Module_basic",
"ElementwiseBitwiseXorModule_basic",
"ElementwiseBitwiseXorStaticShapeModule_basic",
"ElementwiseClampMaxModule_basic",
"ElementwiseClampMinModule_basic",
"ElementwiseClampModule_basic",
"ElementwiseClampTensorInt8Module_basic",
"ElementwiseCosIntModule_basic",
"ElementwiseCosModule_basic",
"ElementwiseCoshIntModule_basic",
"ElementwiseCoshModule_basic",
"ElementwiseDequantizePerChannelModule_basic",
"ElementwiseDequantizePerTensorModule_basic",
"ElementwiseDivScalarRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivScalarRoundingModeTruncModule_basic",
"ElementwiseDivScalarRoundingModeTruncStaticModule_basic",
"ElementwiseDivTensorFloatModule_basic",
"ElementwiseDivTensorIntegerModule_basic",
"ElementwiseDivTensorRoundingModeFloorIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeFloorModule_basic",
"ElementwiseDivTensorRoundingModeTruncIntStaticModule_basic",
"ElementwiseDivTensorRoundingModeTruncModule_basic",
"ElementwiseDivTensorRoundingModeTruncStaticModule_basic",
"ElementwiseDivTensorUnsignedIntegerModule_basic",
"ElementwiseEluNonDefaultModule_basic",
"ElementwiseEqBoolScalarModule_basic",
"ElementwiseEqDiffWidthScalarModule_basic",
"ElementwiseErfIntModule_basic",
"ElementwiseErfModule_basic",
"ElementwiseExpIntModule_basic",
"ElementwiseExpm1IntModule_basic",
"ElementwiseExpm1Module_basic",
"ElementwiseFlattenBroadcastModule_basic",
[Linalg] Promote type for compare tensor op (#3416)
2024-06-05 07:05:39 +08:00
"ElementwiseFloatTensorGtIntTensorModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"ElementwiseFmodTensor_Float_basic",
"ElementwiseFmodTensor_Int_Float_basic",
"ElementwiseFmodTensor_Int_basic",
"ElementwiseGeFloatIntScalarModule_basic",
"ElementwiseGeFloatScalarModule_basic",
"ElementwiseGeFloatTensorModule_basic",
"ElementwiseGeIntScalarModule_basic",
"ElementwiseGeIntTensorModule_basic",
"ElementwiseGeMixedIntScalarModule_basic",
"ElementwiseGtMixed2ScalarModule_basic",
"ElementwiseIntTensorLtFloatScalarModule_basic",
"ElementwiseIsinfModule_basic",
"ElementwiseLeFloatTensorNanModule_basic",
"ElementwiseLeMixedIntScalarModule_basic",
"ElementwiseLog10IntModule_basic",
"ElementwiseLog2IntModule_basic",
"ElementwiseLogIntModule_basic",
"ElementwiseLtDiffWidthScalarModule_basic",
"ElementwiseMishModule_basic",
"ElementwiseMulScalarModule_basic",
"ElementwiseMulTensorComplexDiffModule_basic",
"ElementwiseMulTensorComplexModule_basic",
"ElementwiseMulTensorFloatModule_basic",
"ElementwiseMulTensorIntModule_basic",
"ElementwiseNanToNumModule_Basic",
"ElementwiseOrTensorModule_basic",
"ElementwiseOrTensorStaticShapeModule_basic",
"ElementwisePowModule_basic",
"ElementwisePowScalarModule_basic",
"ElementwisePowTensorBroadcastModule_basic",
"ElementwisePowTensorBroadcastStaticModule_basic",
"ElementwisePowTensorModule_basic",
"ElementwisePowTensorStaticModule_basic",
"ElementwiseQuantizePerTensorModule_basic",
"ElementwiseQuantizePerTensorUIntModule_basic",
"ElementwiseReciprocalIntModule_basic",
"ElementwiseRelu6Module_basic",
"ElementwiseRemainderScalarModule_Bool_basic",
"ElementwiseRemainderScalarModule_Int_Float_basic",
"ElementwiseRemainderScalarModule_Int_basic",
"ElementwiseRemainderTensorModule_Int_Float_basic",
"ElementwiseRemainderTensorModule_Int_basic",
"ElementwiseRsqrtIntModule_basic",
"ElementwiseSgnModule_basic",
"ElementwiseSigmoidIntModule_basic",
"ElementwiseSinIntModule_basic",
"ElementwiseSinModule_basic",
"ElementwiseSinhIntModule_basic",
"ElementwiseSinhModule_basic",
"ElementwiseSqrtIntModule_basic",
"ElementwiseSubScalarIntModule_basic",
"ElementwiseTanIntModule_basic",
"ElementwiseTanModule_basic",
"ElementwiseTernaryModule_basic",
"ElementwiseToDtypeF32ToI64Module_basic",
"ElementwiseToDtypeI64ToI8Module_basic",
"ElementwiseToDtypeI64ToUI8Module_basic",
"ElementwiseTruncIntModule_basic",
"ElementwiseTruncModule_basic",
"ElementwiseUnaryIntModule_basic",
"ElementwiseUnsqueezeNegDimsModule_basic",
"ElementwiseWhereScalarOtherModule_basic",
"ElementwiseWhereScalarSelfModule_basic",
"ElementwiseWhereSelfModule_basic",
"EmbeddingModule1DIndices_basic",
"EmbeddingModuleF16_basic",
"EmbeddingModuleI32Static_basic",
"EmbeddingModuleI32_basic",
"EmbeddingModuleI64_basic",
"EmptyLikeMemoryFormatModule_basic",
"EmptyLikeModule_defaultDtype",
"EmptyLikeModule_falsePinMemory",
"EmptyLikeModule_float",
"EmptyLikeModule_int",
"EmptyStridedModule_basic",
"EmptyStridedSizeIntStrideModule_basic",
"EqIntModule_basic",
"ExpandAsFloatModule_basic",
"ExpandAsIntModule_basic",
"ExpandModule_basic",
"ExponentialModule_basic",
"FakeQuantizePerTensorAffineDynamicShapeModule_basic",
"FakeQuantizePerTensorAffineModule_basic",
"FakeQuantizePerTensorAffineRoundToEvenModule_basic",
"Fill_TensorFloat32WithFloat32_basic",
"Fill_TensorFloat32WithFloat64_basic",
"Fill_TensorFloat32WithInt64_basic",
"Fill_TensorFloat64WithFloat32_basic",
"Fill_TensorFloat64WithFloat64_basic",
"Fill_TensorFloat64WithInt64_basic",
"FlattenDynamicModuleCollapseAll_basic",
"FlattenDynamicModule_basic",
"FlattenRank0Module_basic",
"FlipModuleStaticShape_basic",
"FlipModule_basic",
"FlipNegativeIndexModule_basic",
"FloatImplicitModule_basic",
"FullLikeModuleDefaultDtype_basic",
"FullLikeModuleFalsePinMemory_basic",
"FullLikeModuleFloat2D_basic",
"FullLikeModuleFloat3D_basic",
"FullLikeModuleInt2D_basic",
"FullLikeModuleInt3D_basic",
"Gather2DInputModdule_basic",
"GatherModule_basic",
"GatherNegativeDimModule_basic",
"GatherRandomIndexModule_basic",
"GeFloatIntModule_basic",
"GeFloatModule_basic",
"GeIntModule_basic",
"GeluBackwardModule_basic",
"GridSamplerBasic1_basic",
"GridSamplerBasic2_basic",
"GridSamplerBasic3_basic",
"GridSamplerBasic4_basic",
"GtFloatIntModule_basic",
"GtIntModule_basic",
"HBC_basic",
"HardTanhIntModule_basic",
"HardTanhModule_basic",
"HardtanhBackward_basic",
"IndexPut1DFloatAccumulateModule_basic",
"IndexPut1DFloatNonAccumulateModule_basic",
"IndexPut1DIntAccumulateModule_basic",
"IndexPut1DIntNonAccumulateModule_basic",
"IndexPut2DFloatAccumulateModule_basic",
"IndexPut2DFloatNonAccumulateModule_basic",
"IndexPut2DIntAccumulateModule_basic",
"IndexPut2DIntNonAccumulateModule_basic",
"IndexPut3DFloatAccumulateModule_basic",
"IndexPut3DFloatNonAccumulateModule_basic",
"IndexPut3DIntAccumulateModule_basic",
"IndexPut3DIntNonAccumulateModule_basic",
"IndexPutHackedTwin1DFloatAccumulateModule_basic",
"IndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin1DIntAccumulateModule_basic",
"IndexPutHackedTwin1DIntNonAccumulateModule_basic",
"IndexPutHackedTwin2DFloatAccumulateModule_basic",
"IndexPutHackedTwin2DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin2DIntAccumulateModule_basic",
"IndexPutHackedTwin2DIntNonAccumulateModule_basic",
"IndexPutHackedTwin3DFloatAccumulateModule_basic",
"IndexPutHackedTwin3DFloatNonAccumulateModule_basic",
"IndexPutHackedTwin3DIntAccumulateModule_basic",
"IndexPutHackedTwin3DIntNonAccumulateModule_basic",
"IndexPutImpl1DFloatAccumulateModule_basic",
"IndexPutImpl1DFloatNonAccumulateModule_basic",
"IndexPutImpl1DIntAccumulateModule_basic",
"IndexPutImpl1DIntNonAccumulateModule_basic",
"IndexPutImpl2DFloatAccumulateModule_basic",
"IndexPutImpl2DFloatNonAccumulateModule_basic",
"IndexPutImpl2DImplicitModule_basic",
"IndexPutImpl2DIndexModule_basic",
"IndexPutImpl2DNoneIndexStaticModule_basic",
"IndexPutImpl3DFloatAccumulateModule_basic",
"IndexPutImpl3DFloatNonAccumulateModule_basic",
"IndexPutImplIndexWithNoneModule_basic",
"IndexSelectDynamicIndexSizeModule_basic",
"IndexSelectDynamicInputSizeModule_basic",
"IndexSelectDynamicModulebasic",
"IndexSelectNegativeDimModule_basic",
"IndexSelectRank0IdxModule_basic",
"IndexSelectSingleIdxModule_basic",
"IndexSelectTwoIdxModule_basic",
"IndexSelectWholeDimensionModule_basic",
"IndexSelectWholeTensorModule_basic",
"IndexTensorDyanmicInputContiguousWithNoneModule_basic",
"IndexTensorDyanmicInputNonContiguousWithNoneModule_basic",
"IndexTensorHackedTwinModule3dInput_basic",
"IndexTensorHackedTwinModule_basic",
"IndexTensorHackedTwinMultiInputNonContiguousMultipleStaticDims_basic",
"IndexTensorModule3dInput_basic",
"IndexTensorModule_basic",
"IndexTensorMultiIndexStaticModule_basic",
"IndexTensorMultiInputContiguousCenter_basic",
"IndexTensorMultiInputContiguousOneDimDynamic_basic",
"IndexTensorMultiInputNonContiguousDynamic_basic",
"IndexTensorMultiInputNonContiguousMultipleStaticDims_basic",
"IndexTensorMultiInputNonContiguousOneDimDynamic_basic",
"IndexTensorMultiInputNonContiguous_basic",
"IndexTensorMultiInputOneDim_basic",
"IndexTensorMultiInputThreeIndexers_basic",
"IndexTensorMultiInput_basic",
"IndexTensorNegativeIndexModule_basic",
"IndexTensorSelectDimModule_basic",
"IndexTensorStaticContiguousWithNoneModule_basic",
"IndexTensorStaticModule_basic",
"IndexTensorStaticNonContiguousWithNoneModule_basic",
Modifies onnx resize lowering to fix numerical issues (#3381) Updates: - some unsupported modes are now going to report a match failure for unsupported coordinate transformation modes. - fixes a bug that was introduced in the last patch for resize (my bad...) - uses actual x and y coordinates for computing weights in bilinear interpolation (rather than eps modified values) - slightly simplifies the bilinear interpolation payload for readability and performance - passes coordinate transformation mode information from an onnx.Resize op to the mode string for the aten._interpolate op. This allows us to perform custom logic in the torch->linalg lowering to support onnx.Resize options without losing the default behaviors of the interpolate op.
2024-05-31 08:34:37 +08:00
"InterpolateDynamicModule_sizes_bilinear",
"InterpolateDynamicModule_sizes_nearest",
"InterpolateDynamicModule_scales_recompute_bilinear",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"IntFloatModule_basic",
"IntImplicitModule_basic",
"IouOfModule_basic",
"IsFloatingPointFloat_True",
"IsFloatingPointInt_False",
"IscloseStaticModuleTrue_basic",
"IscloseStaticModule_basic",
"LeakyReluBackwardModule_basic",
"LeakyReluBackwardStaticModule_basic",
"LenStrModule_basic",
"LiftFreshCopyModule_basic",
"LinalgNormKeepDimComplexModule_basic",
"LinalgNormModule_basic",
"LinalgVectorNormComplexModule_basic",
"LinalgVectorNormKeepDimModule_basic",
"LinalgVectorNormModule_basic",
"LogSoftmaxBackwardModule_basic",
"LogSoftmaxIntModule_basic",
"MaskedFillTensorFloatValueModule_basic",
"MatmulBroadcastBatchDim_basic",
"MatmulSingleDynamicBatchDim_basic",
"Matmul_2d",
"Matmul_4d",
"Matmul_matvec",
"Matmul_vecmat",
"MaxPool1dCeilModeTrueModule_basic",
"MaxPool1dEmptyStrideStaticModule_basic",
"MaxPool1dModule_basic",
"MaxPool1dStaticCeilModeTrueModule_basic",
"MaxPool1dStaticModule_basic",
"MaxPool2dCeilModeTrueModule_basic",
"MaxPool2dModule_basic",
"MaxPool2dWithIndicesAllNegativeValuesModule_basic",
"MaxPool2dWithIndicesAllOnesModule_basic",
"MaxPool2dWithIndicesBackwardDynamic3DModule_basic",
"MaxPool2dWithIndicesBackwardDynamic4DModule_basic",
"MaxPool2dWithIndicesBackwardStatic3DModule_basic",
"MaxPool2dWithIndicesBackwardStatic4DModule_basic",
"MaxPool2dWithIndicesCeilModeTrueModule_basic",
"MaxPool2dWithIndicesFullSizeKernelModule_basic",
"MaxPool2dWithIndicesModule_basic",
"MaxPool2dWithIndicesNonDefaultDilationModule_basic",
"MaxPool2dWithIndicesNonDefaultPaddingModule_basic",
"MaxPool2dWithIndicesNonDefaultParamsModule_basic",
"MaxPool2dWithIndicesNonDefaultStrideModule_basic",
"MaxPool2dWithIndicesStaticModule_basic",
"MaxPool3dCeilModeTrueModule_basic",
"MaxPool3dEmptyStrideStaticModule_basic",
"MaxPool3dLargeDatadModule_basic",
"MaxPool3dModuleRandomSimple_basic",
"MaxPool3dModule_basic",
"MaxPool3dStaticCeilModeTrueModule_basic",
"MaxPool3dStaticModule_basic",
[TorchToLinalg] Support lowering MaxPool3dWithIndices (#3652) Support torch.MaxPool3dWithIndices lowering to linalg backend.
2024-08-28 03:14:25 +08:00
"MaxPool3dWithIndicesAllNegativeValuesModule_basic",
"MaxPool3dWithIndicesAllOnesModule_basic",
"MaxPool3dWithIndicesCeilModeTrueModule_basic",
"MaxPool3dWithIndicesFullSizeKernelModule_basic",
"MaxPool3dWithIndicesModule_basic",
"MaxPool3dWithIndicesNonDefaultDilationModule_basic",
"MaxPool3dWithIndicesNonDefaultPaddingModule_basic",
"MaxPool3dWithIndicesNonDefaultParamsModule_basic",
"MaxPool3dWithIndicesNonDefaultStrideModule_basic",
"MaxPool3dWithIndicesStaticModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"MeanDimAllReduceKeepdimModule_basic",
"MeanDimAllReduceModule_basic",
"MeanDimDtypeModule_basic",
"MeanDimEmptyDimModule_basic",
"MeanDimKeepdimModule_basic",
"MeanDimModule_basic",
"MeanDimNegativeModule_basic",
"MeanDimNoneDimModule_basic",
"MeanDtypeModule_basic",
"MeanDynamicSizesModule_basic",
"MeanModule_basic",
"Mlp1LayerModule_basic",
"Mlp2LayerModuleNoBias_basic",
"Mlp2LayerModule_basic",
"MmModule_basic",
"MmModule_chained",
"MmTanhModule_basic",
"MobilenetV3Module_basic",
"MoveDimIntNegativeIndexModule_basic",
"MseLossMeanReductionModule_basic",
"MseLossSumReductionWithDifferentElemTypeModule_basic",
"MulFloatModule_basic",
"MulIntModule_basic",
"Mv_basic",
"NarrowHorizontalTest2_basic",
"NarrowHorizontalTest_basic",
"NarrowTensorHorizontalModule_basic",
"NarrowTensorVerticalModule_basic",
"NarrowVerticalTest2_basic",
"NarrowVerticalTest_basic",
"NativeBatchNorm1DModule_basic",
"NativeBatchNorm2DModule_basic",
"NativeBatchNorm3DModule_basic",
"NativeBatchNormNoneWeightModule_basic",
"NativeDropoutEvalFloatModule_basic",
"NativeDropoutTrainModule_basic",
"NativeDropoutTrainStaticShapeModule_basic",
"NativeGroupNormBackwardModule_basic",
"NativeGroupNormModule_basic",
"NativeLayerNormDynamicModule_basic",
"NativeLayerNormModule4D_basic",
"NativeLayerNormModule_basic",
"NeFloatIntModule_basic",
"NeIntModule_basic",
"NewEmptyStridedModuleDefaultDtype_basic",
"NllLossModuleBackward1DMeanWeight_basic",
"NllLossModuleBackward1DMean_basic",
"NllLossModuleBackward1DSumWeight_basic",
"NllLossModuleBackward1DSum_basic",
"NllLossModuleBackward1DWeight_basic",
"NllLossModuleBackward1D_basic",
"NllLossModuleBackwardMeanWeight_basic",
"NllLossModuleBackwardMean_basic",
"NllLossModuleBackwardSumWeight_basic",
"NllLossModuleBackwardSum_basic",
"NllLossModuleBackwardWeight_basic",
"NllLossModuleBackward_basic",
"NllLossModuleBackward_ignore_index",
"NllLossModule_1D_basic",
"NllLossModule_basic",
"NllLossModule_ignore_index_out_of_bounds_basic",
"NllLossModule_mean_basic",
"NllLossModule_sum_basic",
"NormScalarComplexModule_basic",
"NormScalarModule_basic",
"NormScalarOptDimKeepDimComplexModule_basic",
"NormScalarOptDimKeepDimModule_basic",
"NormScalarOptDimModule_basic",
"NormalFunctionalModule_basic",
"NormalizeModule_basic",
"NumToTensorFloatModule_basic",
"NumToTensorIntModule_basic",
"NumelModule_basic",
"NumelZeroRankModule_basic",
"OneHotModule_basic",
"OnesLikeModule_defaultDtype",
"OnesLikeModule_falsePinMemory",
"OnesLikeModule_float",
"OnesLikeModule_int",
"PadModule_basic",
"PadWithNoneValModule_basic",
"PermuteNegativeIndexModule_basic",
"PixelShuffleModuleFullDynamic_basic",
"PixelShuffleModuleSpatiallyDynamic_basic",
"PixelShuffleModuleSpatiallyStatic_basic",
"PixelShuffleModuleStaticRank3Int64_basic",
"PixelShuffleModuleStaticRank4Float32_basic",
"PowIntFloatModule_basic",
"PrimMaxIntModule_basic",
"PrimMinIntDynamicModule_basic",
"PrimMinIntModule_basic",
"PrimsConvertElementTypeModule_basic",
"PrimsIotaModule_basic",
"PrimsSqueezeEmptyDimensionsModule_basic",
"PrimsSqueezeModule_basic",
"PrimsViewOfModule_basic",
"PrimsViewOfZeroRankModule_basic",
"QuantizedBatchedInputSingleLayer_basic",
"QuantizedMLP_basic",
"QuantizedNoLayer_basic",
"QuantizedReluInt32_basic",
"QuantizedReluInt8_basic",
"QuantizedReluUint8_basic",
"QuantizedSingleLayer_basic",
"RandIntDtypeModule_basic",
"RandIntLowDtypeModule_basic",
"RandIntLowModule_basic",
"RandIntModule_basic",
"RandIntPinMemoryModule_basic",
"RandLikeDtypeModule_basic",
"RandLikeModule_basic",
"RandModule_basic",
"RandnDtypeDeviceModule_basic",
"RandnGeneratorF64Module_basic",
"RandnGeneratorModule_basic",
"RandnLikeDtypeModule_basic",
"RandnLikeModule_basic",
"RandnModule_basic",
"ReduceAllBoolModule_basic",
"ReduceAllDimBool_basic",
"ReduceAllDimEmpty_basic",
"ReduceAllDimFloat_basic",
"ReduceAllDimInt_basic",
"ReduceAllFloatModule_basic",
"ReduceAllIntModule_basic",
"ReduceAmaxKeepDim_basic",
"ReduceAmaxMultiDim_basic",
"ReduceAmaxOutOfOrderDim_basic",
"ReduceAmaxSingleDim_basic",
"ReduceAnyBoolModule_basic",
"ReduceAnyFloatModule_basic",
"ReduceAnyIntModule_basic",
"ReduceFrobeniusNormComplexModule_basic",
"ReduceL1NormComplexModule_basic",
"ReduceL1NormModule_basic",
"ReduceL1NormWithDTypeModule_basic",
"ReduceL2NormComplexModule_basic",
"ReduceL2NormModule_basic",
"ReduceL3NormAllDimsModule_basic",
"ReduceL3NormKeepDimComplexModule_basic",
"ReduceL3NormKeepDimModule_basic",
"ReduceLN3NormModule_basic",
"ReduceMaxAllDims_basic",
"ReduceMaxAlongDimNegative_basic",
"ReduceMaxAlongDimSignedInt_basic",
"ReduceMaxAlongDimUnsignedInt_basic",
"ReduceMaxAlongDim_basic",
"ReduceMaxFloatModule_basic",
"ReduceMaxKeepDimReturnBoth_basic",
"ReduceMaxKeepDim_basic",
"ReduceMaxNegativeDim_basic",
"ReduceMaxSignedIntModule_basic",
"ReduceMaxUnsignedIntModule_basic",
"ReduceMinAlongDimNegative_basic",
"ReduceMinAlongDimSignedInt_basic",
"ReduceMinAlongDimUnsignedInt_basic",
"ReduceMinAlongDim_basic",
"ReduceMinFloatModule_basic",
"ReduceMinKeepDimReturnBoth_basic",
"ReduceMinKeepDim_basic",
"ReduceMinSignedIntModule_basic",
"ReduceMinUnsignedIntModule_basic",
"ReduceProdDimIntFloatModule_basic",
"ReduceProdDtypeFloatModule_basic",
"ReduceProdDtypeIntModule_basic",
"ReduceProdElementTypeBoolModule_basic",
"ReduceProdFloatModule_basic",
"ReduceProdSignedIntModule_basic",
"ReduceProdUnsignedIntModule_basic",
"ReduceSumDimIntListDtypeFloatModule_basic",
"ReduceSumDimIntListDtypeIntModule_basic",
"ReduceSumDimIntListElementTypeBoolModule_basic",
"ReduceSumDimIntListEmptyDimModule_basic",
"ReduceSumDtypeFloatModule_basic",
"ReduceSumDtypeIntModule_basic",
"ReduceSumElementTypeBoolModule_basic",
"ReduceSumFloatModule_basic",
"ReduceSumSignedIntModule_basic",
"ReduceSumUnsignedIntModule_basic",
"ReflectionPad1dModule2dInput_Right",
"ReflectionPad1dModule2dInput_basic",
"ReflectionPad1dModule3dInput_Left",
"ReflectionPad1dModule3dInput_basic",
"ReflectionPad2dModule_Bottom",
"ReflectionPad2dModule_Left",
"ReflectionPad2dModule_Right",
"ReflectionPad2dModule_Top",
"ReflectionPad2dModule_basic",
"RepeatModule_basic",
"ReplicationPad2dModule_basic",
"ReplicationPad2dModule_bottom0",
"ReplicationPad2dModule_left0",
"ReplicationPad2dModule_right0",
"ReplicationPad2dModule_top0",
"ResNet18Module_basic",
"ResNet18StaticModule_basic",
"ReshapeAliasCollapseModule_basic",
"ReshapeAliasExpandModule_basic",
"ReshapeCollapseModule_basic",
"ReshapeDynamicModule_basic",
"ReshapeExpandModule_basic",
"RollModule_basic",
"RsubIntModule_noalpha_basic",
"ScalarConstantTupleModule_basic",
"ScalarImplicitFloatModule_basic",
"ScalarImplicitIntModule_basic",
[Torch] [TMTensor] Added mask and is_causal support for torch.aten.scaled_dot_product_attention (#3690) Enabled mask and is_causal parameters for torch.aten.scaled_dot_product attention + relevant comments + tests. The tests added highlight the new capabilities introduced in this PR, including: Attention with F16 mask Attention with Boolean mask Causal attention with same Q K V shapes Causal attention without Q K V shapes Made sure that one cannot input both mask and is_causal.
2024-09-10 06:51:41 +08:00
# REMOVE WHEN ENABLE_GQA IS ADDED
"ScaledDotProductAttentionBoolMaskModule_basic",
"ScaledDotProductAttentionDifferentCausalModule_basic",
"ScaledDotProductAttentionSameCausalModule_basic",
"ScaledDotProductAttentionSameDynamicModule_basic",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"ScatterReduceFloatMaxModule",
"ScatterReduceFloatMaxModuleIncludeSelf",
"ScatterReduceFloatMeanModule",
"ScatterReduceFloatMeanModuleIncludeSelf",
"ScatterReduceFloatMinModule",
"ScatterReduceFloatMinModuleIncludeSelf",
"ScatterReduceFloatProdModule",
"ScatterReduceFloatProdModuleIncludeSelf",
"ScatterReduceFloatSumModule",
"ScatterReduceFloatSumModuleIncludeSelf",
"ScatterReduceIntMaxModule",
"ScatterReduceIntMaxModuleIncludeSelf",
"ScatterReduceIntMeanModule",
"ScatterReduceIntMeanModuleIncludeSelf",
"ScatterReduceIntMinModule",
"ScatterReduceIntMinModuleIncludeSelf",
"ScatterReduceIntProdModule",
"ScatterReduceIntProdModuleIncludeSelf",
"ScatterReduceIntSumModule",
"ScatterReduceIntSumModuleIncludeSelf",
"ScatterSrcModule_basic",
"ScatterSrcStaticModule_basic",
"ScatterValueFloatModule_basic",
"ScatterValueIntModule_basic",
"SelectIntModule_basic",
"SelectIntNegativeDimAndIndexStaticModule_basic",
"SelectScattertModule_basic",
"SelectScattertStaticModule_basic",
implement lowering of torch.aten._linalg_slogdet (#3524)
2024-07-19 13:54:43 +08:00
"SignAndLogarithmOfDeterminantModule_F32",
"SignAndLogarithmOfDeterminantBatchedModule_F32",
"SignAndLogarithmOfDeterminantDynamicModule_F32",
[ONNX][TOSA] Adds ONNX to TOSA e2e tests (#3358) - Refactors OnnxBackend to be generic and consume any Torch backend. --------- Signed-off-by: Suraj Sudhir <suraj.sudhir@arm.com>
2024-05-17 12:44:26 +08:00
"SliceCopyEndGreaterThanDimSize_Module_basic",
"SliceCopyNegative_Module_basic",
"SliceCopyNonZeroDim_Module_basic",
"SliceCopy_Module_basic",
"SliceEndSleStartModule_basic",
"SliceModule_basic",
"SliceNegIdxModule_basic",
"SliceOutOfLowerBoundEndIndexModule_basic",
"SliceOutOfLowerBoundStartIndexModule_basic",
"SliceOutOfUpperBoundIndexModule_basic",
"SliceScatterModule_basic",
"SliceScatterNegativeDimModule_basic",
"SliceScatterNegativeEndModule_basic",
"SliceScatterStaticModule_basic",
"SliceScatterStepVariationModule_basic",
"SliceScatterZeroDimModule_basic",
"SliceSingleIdxModule_basic",
"SliceSizeTwoStepModule_basic",
"SliceStartEqEndModule_basic",
"SoftmaxBackwardModule_basic",
"SoftmaxIntArgTypeF64Module_basic",
"SoftmaxIntModule_basic",
"SoftmaxIntNegDimModule_basic",
"SoftmaxIntNonNoneDtypeModule_basic",
"SoftplusModule_basic",
"SortIntListReverse_basic",
"SortIntList_basic",
"SortTensorDescending_basic",
"SortTensorInteger_basic",
"SortTensorNegativeDimension_basic",
"SortTensorSpecificDimension_basic",
"SortTensor_basic",
"SplitDimDynamicModule_basic",
"SplitDimStaticModule_basic",
"SplitWithSizes_Module_basic",
"SqrtIntConstantModule_basic",
"SqrtIntModule_basic",
"SqueezeDimModule_dynamic",
"SqueezeDimModule_negDim",
"StdBiasedModule_basic",
"StdCorrectionAllDimReduceModule_basic",
"StdCorrectionEmptyDimModule_basic",
"StdCorrectionKeepDimModule_basic",
"StdCorrectionLargeInputModule_basic",
"StdCorrectionModule_basic",
"StdCorrectionNoneModule_basic",
"StdCorrectionSingleDimReduceModule_basic",
"StdDimBiasedModule_basic",
"StdDimEmptyDimModule_basic",
"StdDimKeepDimFalseModule_basic",
"StdDimKeepDimTrueModule_basic",
"StdDimNoneDimModule_basic",
"StdUnbiasedModule_basic",
"SubFloatModule_basic",
"SubIntModule_basic",
"TanhBackward_basic",
"TensorToBoolZeroRank_basic",
"TensorToBool_basic",
"TensorToFloatZeroRank_basic",
"TensorToFloat_basic",
"TensorToIntZeroRank_basic",
"TensorToInt_basic",
"TensorsConcatModule_basic",
"TensorsConcatNegativeDimModule_basic",
"TensorsConcatPromoteDTypeModule_basic",
"TensorsStackModule_basic",
"TensorsStackNegativeDimModule_basic",
"TensorsStackPromoteDTypeModule_basic",
"TensorsStackSingleElementListModule_basic",
"TestMultipleTensorAndPrimitiveTypesReturn_basic",
"Threshold1dFloatModule_basic",
"Threshold1dIntI32Module_basic",
"Threshold1dIntModule_basic",
"Threshold2dFloatModule_basic",
"Threshold2dIntModule_basic",
"Threshold3dFloatModule_basic",
"Threshold3dIntModule_basic",
"ThresholdBackward1dFloatModule_basic",
"ThresholdBackward1dIntModule_basic",
"ThresholdBackward1dMixedModule_basic",
"ThresholdBackward2dFloatModule_basic",
"ThresholdBackward2dIntModule_basic",
"ThresholdBackward2dMixedModule_basic",
"ThresholdBackward3dFloatModule_basic",
"ThresholdBackward3dIntModule_basic",
"ThresholdBackward3dMixedModule_basic",
"TileBigDimsSizeModule_basic",
"TileSmallDimsSizeModule_basic",
"ToCopyBoolDTypeStaticModule_basic",
"ToCopyModule_basic",
"ToCopyWithDTypeFalsePinMemoryModule_basic",
"ToCopyWithDTypeModule_basic",
"ToDtypeLayoutCPUModule_basic",
"ToDtypeLayoutNoneModule_basic",
"ToDtypeLayoutStridedModule_basic",
"TorchPrimLoopForLikeModule_basic",
"TorchPrimLoopWhileLikeModule_basic",
"TraceModule_basic",
"TraceModule_empty",
"TraceModule_nonsquare",
"TraceSignedIntModule_basic",
"TraceUnsignedIntModule_basic",
"TraceUnsignedIntModule_empty",
"TriuBroadcastModule_basic",
"TriuModule_basic",
"TupleModule_basic",
"TypeAsDifferentModule_basic",
"TypeConversionF32ToF64Module_basic",
"TypeConversionF64ToF32Module_basic",
"TypeConversionI1ToF32Module_basic",
"TypeConversionI1ToF64Module_basic",
"TypeConversionI1ToI32Module_basic",
"TypeConversionI1ToI64Module_basic",
"TypeConversionI32ToI64Module_basic",
"TypeConversionI64ToI32Module_basic",
"TypePromotionDifferentCategoryModule_basic",
"TypePromotionSameCategoryDifferentWidthModule_basic",
"TypePromotionZeroRankHigherCategoryModule_basic",
"UnflattenIntNegativeOneDimStaticModule_basic",
"UnflattenIntNegativeOneSizeStaticModule_basic",
"UnflattenIntStaticModule_basic",
"UnflattenStaticModule_basic",
"UniformModule_basic",
"UniformNoCorrelationModule_basic",
"UniformStaticShapeModule_basic",
"UnsafeIndexPutHackedTwin1DFloatNonAccumulateModule_basic",
"UnsafeView1DFoldModule_basic",
"UnsafeViewCollapseDynamicWithAtenSizeIntModule_basic",
"UnsafeViewCollapseModule_basic",
"UnsafeViewDynamicExpandModule_basic",
"UnsafeViewDynamicExpandWithAtenSizeIntModule_basic",
"UnsafeViewExpandModule_basic",
"UpSampleNearest2dBackwardScalesNone_basic",
"UpSampleNearest2dBackward_basic",
"UpSampleNearest2dDynamicFactor_basic",
"UpSampleNearest2dDynamicSize_basic",
"UpSampleNearest2dStaticFactor_basic",
"UpSampleNearest2dStaticSize_basic",
"UpSampleNearest2d_basic",
"VarBiasedModule_basic",
"VarCorrectionAllDimReduceModule_basic",
"VarCorrectionEmptyDimModule_basic",
"VarCorrectionKeepDimModule_basic",
"VarCorrectionLargeInputModule_basic",
"VarCorrectionModule_basic",
"VarCorrectionNoneModule_basic",
"VarCorrectionSingleDimReduceModule_basic",
"VarDimAllDimReduceModule_basic",
"VarDimBiasedModule_basic",
"VarDimEmptyDimModule_basic",
"VarDimModule_basic",
"VarDimMultiDimModule_basic",
"VarDimNegativeModule_basic",
"VarDimNoneDimModule_basic",
"VarDimSingleDimModule_basic",
"VarDimUnbiasedModule_basic",
"VarMeanBiasedModule_basic",
"VarMeanCorrectionModule_basic",
"VarMeanCorrectionNoneModule_basic",
"VarMeanDimBiasedModule_basic",
"VarMeanDimModule_basic",
"VarMeanUnbiasedModule_basic",
"VarUnbiasedModule_basic",
"View1DFoldModule_basic",
"ViewCollapseDynamicWithAtenSizeIntModule_basic",
"ViewCollapseModule_basic",
"ViewDynamicExpandCollapseModule_basic",
"ViewDynamicExpandCollapseWithAtenIntModule_basic",
"ViewDynamicExpandCollapseWithParallelUnknownDimModule_basic",
"ViewDynamicExpandModule_basic",
"ViewDynamicExpandWithAtenSizeIntModule_basic",
"ViewExpandDynamicDimModule_basic",
"ViewFlattenAndExpandModule_basic",
"ViewNoChange1dModule_basic",
"ViewNoChange2dModule_basic",
"ViewNoChange3dModule_basic",
"ViewSizeDimFollowedByCollapsedOnesModule_basic",
"ViewSizeDimFollowedByExpandedOnesModule_basic",
"ViewSizeDimLedAndFollowedByCollapsedOnesModule_basic",
"ViewSizeDimLedAndFollowedByExpandedOnesModule_basic",
"ViewSizeDimLedByCollapsedOnesModule_basic",
"ViewSizeDimLedByExpandedOnesModule_basic",
"ViewSizeFromOtherTensor_basic",
"ZeroFloat32Module_basic",
"ZeroInt32Module_basic",
"ZeroInt64Module_basic",
"ZerosLikeModule_defaultDtype",
"ZerosLikeModule_falsePinMemory",
"ZerosLikeModule_float",
"ZerosLikeModule_int",
"_Convolution2DAllFalseModule_basic",
"_Convolution2DBenchmarkModule_basic",
"_Convolution2DCudnnModule_basic",
"_Convolution2DDeterministicModule_basic",
"_Convolution2DTF32Module_basic",
"_ConvolutionDeprecated2DAllFalseModule_basic",
"_ConvolutionDeprecated2DBenchmarkModule_basic",
"_ConvolutionDeprecated2DCudnnModule_basic",
"_ConvolutionDeprecated2DDeterministicModule_basic",
"_LogSoftmaxModule_basic",
"_SoftmaxModule_basic",
}