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

[cmake] Torch-MLIR doesn't build with CMake setting 'BUILD_SHARED_LIBS=ON'

christopherbate opened this issue 9 days ago

Update C onnx importer

giacs-epic opened this pull request 10 days ago

Clarify `min_val` semantics for `torch.symbolic_int` op

sjain-stanford opened this pull request 10 days ago

[TOSA] Add acc_type to TOSA conv ops in Torch to TOSA legalization

justin-ngo-arm opened this pull request 10 days ago

[TOSA] Update TOSA pad shape in Torch to TOSA legalization

justin-ngo-arm opened this pull request 10 days ago

Backend-legal-ops argument for fx lowering

Abhishek-TyRnT opened this pull request 11 days ago

[ONNX] improve regex matching in onnx-importer name sanitization

zjgarvey opened this pull request 11 days ago

support decomposition of aten.adaptive_max_pool2d

qingyunqu opened this pull request 12 days ago

[QUERY] How to stop torch-mlir from decomposing certain ops when using torch dynamo

Abhishek-TyRnT opened this issue 14 days ago

Rahuls/ltc debug

rahuls-cerebras opened this pull request 15 days ago

Remove TOSA make_fx configuration

mgehre-amd opened this pull request 16 days ago

[MLIR][TORCH] Add support for `enable_gqa` flag in SDPA op

vivekkhandelwal1 opened this pull request 16 days ago

[TOSA] Add lowering for aten.expm1

justin-ngo-arm opened this pull request 16 days ago

[TOSA] Update tosa.cast check according to TOSA v1.0 spec

justin-ngo-arm opened this pull request 16 days ago

[TMTensor] Cast i1 to i32by extsi instead of trunci for aten scatter_add

AmosLewis opened this pull request 16 days ago

[TOSA] : Fix float to integer cast for `torch.ops.aten.to` lowering.

sahas3 opened this pull request 16 days ago

fix(ONNX): avoids non-scalable dimensions in onnx.resize

bjacobgordon opened this pull request 17 days ago

Lowering program that mutates its inputs - how to lower `torch.copy.to_vtensor` and `torch.overwrite.tensor.contents` ops

dixinzhou opened this issue 17 days ago

[torch-mlir] Support lowering of aten constraint ops

praveen-g-ctt opened this pull request 17 days ago

[ONNX] Delete redundant dynamic dim check for result types

AmosLewis opened this pull request 18 days ago

Use resultElementType for tensor creation for AtenMemoryFormatOp with none dtype.

sahas3 opened this pull request 19 days ago

[ONNX] clarifies error message for upsample interpolation mode

bjacobgordon opened this pull request 21 days ago

[docs] Refresh add_ops.md

bjacobgordon opened this pull request 21 days ago

Semantics of `torch.symbolic_int` `min_value`

kuhar opened this issue 22 days ago

Verbose

rahuls-cerebras opened this pull request 23 days ago

Add verbose options

rahuls-cerebras opened this pull request 25 days ago

Failing to lower torch model to TOSA

kslavka-gsi opened this issue 25 days ago

Update ci.yml

rushins-cerebras opened this pull request 26 days ago

[TOSA] Add reflection_pad3d lowering

sahas3 opened this pull request about 1 month ago

[docs] Refreshes development.md

bjacobgordon opened this pull request about 1 month ago

[Torch] add aten.bilinear op decomposing

dixinzhou opened this pull request about 1 month ago

[torch-mlir][doc] remove MPACT as example

aartbik opened this pull request about 1 month ago

[torch-mlir][sparse] re-enable most sparse test

aartbik opened this pull request about 1 month ago

[torch-mlir] add MPACT as an example torch-mlir based compiler

aartbik opened this pull request about 1 month ago

[cmake] Enable accepting external stablehlo project

shelkesagar29 opened this pull request about 1 month ago

[lib/conversion] Create seed only if needed in `convert-torch-convers…

shelkesagar29 opened this pull request about 1 month ago

[python] Make module imports relative in `fx.py` and `compiler_utils.py`

shelkesagar29 opened this pull request about 1 month ago

[torch] Update `torch.bmm` to use accumulator type

rsuderman opened this pull request about 1 month ago

[ONNX] Add support for asymmetric padding for Onnx.AveragePool op

vivekkhandelwal1 opened this pull request about 1 month ago

[TOSA] Add legalization for torch.aten.unfold

justin-ngo-arm opened this pull request about 1 month ago

Downgrade clang

rahuls-cerebras opened this pull request about 1 month ago

Add attributes support for onnx.nms

jinchen62 opened this pull request about 1 month ago

build: manually update PyTorch version

vivekkhandelwal1 opened this pull request about 1 month ago

[torch-mlir][sparse] register sparse tensor dialect for all rewriting

aartbik opened this pull request about 1 month ago

[ONNX] Add per channel quantization support for Onnx.QLinearConv op

vivekkhandelwal1 opened this pull request about 1 month ago

[ONNX] Fix nonzero output type difference between onnx and torch

AmosLewis opened this pull request about 1 month ago

attention model failed to convert, "error: unknown: unsupported by backend contract: module initializers"

heshuju opened this issue about 2 months ago

[onnx][importer] Merge character sanitization into regex

vinayakdsci opened this pull request about 2 months ago

Revert "[CI] Disable LTC build from CI"

antoniojkim opened this pull request about 2 months ago

memory leak ad-hoc fix

egebeysel opened this pull request about 2 months ago

[CI] Disable LTC build from CI

vivekkhandelwal1 opened this pull request about 2 months ago

Fix LTC build error in CI

vivekkhandelwal1 opened this issue about 2 months ago

[TOSA] Add some more mixed dtype handling

justin-ngo-arm opened this pull request about 2 months ago

Clean up code of onnx.resize lowering

jinchen62 opened this pull request about 2 months ago

Pin and update actions

marbre opened this pull request about 2 months ago

Replace `ubuntu-latest` with specific version

marbre opened this pull request about 2 months ago

Replace unmaintained `create-release` action

marbre opened this pull request about 2 months ago

[MLIR][TORCH] Add support for 1-d group convolution

vivekkhandelwal1 opened this pull request about 2 months ago

Failed to compile onnx.Resize with dynamic shapes

jinchen62 opened this issue about 2 months ago

Wrap indices to be within max range for as_strided + index_select

sahas3 opened this pull request about 2 months ago

[onnx][importer] Sanitize '-' characters in TensorProto names

vinayakdsci opened this pull request about 2 months ago

[demo error] ModuleNotFoundError: No module named 'torch_mlir_e2e_test'

vfdff opened this issue about 2 months ago

build: manually update PyTorch version

vivekkhandelwal1 opened this pull request about 2 months ago

aten.pow / onnx.Pow: Fix (float,int) / (int, float) accuracy

mgehre-amd opened this pull request about 2 months ago

torch.aten._scaled_dot_product_efficient_attention：Failed to legalize operation 'torch.operator' that was explicitly marked illegal

bilu-1 opened this issue about 2 months ago

Add TorchToArith rewrite pattern for AtenGtFloatOp

jinchen62 opened this pull request about 2 months ago

Apply the trait `NoMemoryEffect` to most `ReadOnly` ops

zjgarvey opened this pull request 2 months ago

Fix output size computation for MaxPool2D for ceil_model = true.

sahas3 opened this pull request 2 months ago

address uneccessary dynamic shapes in argmax decomposition

zjgarvey opened this pull request 2 months ago

Support stash_type attribute for onnx.LayerNormalization

jinchen62 opened this pull request 2 months ago

convert to double before float materialization in scalarize shapes

zjgarvey opened this pull request 2 months ago

[TOSA] Add upsample_nearest2d, split_dim, outer, GELU tanh mode and misc

justin-ngo-arm opened this pull request 2 months ago

Assertion when lowering from Torch IR for `AvgPool2d` when `kernel` is a tuple of a single int

sahas3 opened this issue 2 months ago

Llama 3.1 8B F16 Non-Decomposed prefill fails to compile

aviator19941 opened this issue 2 months ago

Add support for the padding variations of conv op

sahas3 opened this pull request 2 months ago

torch_mlir is not a package

bigmover opened this issue 2 months ago

Failing tests for `check-torch-mlir`

mdehling opened this issue 2 months ago

Add `aten.stft.center` and decomposition

giacs-epic opened this pull request 2 months ago

Add bazel configs for some C APIs and pybinds

tkocmathla opened this pull request 2 months ago

[MLIR][TORCH] Add aten.special.expm1 op lowering

vivekkhandelwal1 opened this pull request 2 months ago

[ONNX] reduce min opset versions for Greater and Gather

zjgarvey opened this pull request 2 months ago

[Torch] Add decomposition for 1d torch.nonzero

AmosLewis opened this pull request 2 months ago

Fix a bug for large models in onnx importer.

zjgarvey opened this pull request 2 months ago

[TOSA] Add reflection and replication pad lowering

justin-ngo-arm opened this pull request 2 months ago

Bump LLVM to 813f7c3

vivekkhandelwal1 opened this pull request 2 months ago

[TorchToLinalg] Use `linalg.transpose` instead of `generic` in `permuteTensor`

CoTinker opened this pull request 2 months ago

Support NMS op lowering

jinchen62 opened this pull request 2 months ago

[ONNX] rework some reduction op lowerings

zjgarvey opened this pull request 2 months ago

[ONNX] Remove kernel shape and weight shape equivalence check from Onnx.Conv lowering

vivekkhandelwal1 opened this pull request 2 months ago

Support default padding case for tosa::AvgPool in the presence of count_include_pad

Hanumanth04 opened this pull request 2 months ago

[Torch] support aten.column_stack

yyp0 opened this pull request 2 months ago

failed to legalize operation 'torch.aten.cumsum'

DavidGinten opened this issue 2 months ago

[TorchToStablehlo] support l1_loss, deg2rad, logit

yyp0 opened this pull request 2 months ago

[Torch&stablehlo] support l1_loss, deg2rad, logit lowering

yyp0 opened this pull request 2 months ago

build: manually update PyTorch version

vivekkhandelwal1 opened this pull request 2 months ago

TorchToTosa: ConvertAtenAvgPool2dOp is too restrictive.

dbabokin opened this issue 3 months ago

Add Scalarization Patterns for `AtenToDtypeOp`, `AtenNegOp`, `AtenRemainderTensorOp`

zjgarvey opened this pull request 3 months ago

[TOSA] Add promote type to unary ops and aten.cat lowering

justin-ngo-arm opened this pull request 3 months ago

`Conversion/TorchToTosa/basic.mlir` failing on ARM, due to different random numbers in `torch.aten.uniform`

dbabokin opened this issue 3 months ago

[TorchToLinalg] Issue Tracking for Convolution Support

zjgarvey opened this issue 3 months ago