llvm/torch-mlir

View on GitHub

The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.

Stars: 1.9k
Forks: 736
Open beginner issues: 1
Indexed issues: 373
Avg merge: 5d 22h
Merged PRs (30d): 15

Dominant language: C++
License: No license data
Last GitHub push: Sep 18, 2026
Latest indexed: Sep 19, 2026
Contributing guide: No contributing guide
Code of conduct: No code of conduct
Beginner labels: good first issue help wanted

Beginner friendly Everything

Open All

373 open issues indexed

fx_importer: non-persistent buffers fail with "Could not find state mapping for buffer" Open

Difficulty 2/5 1-3 hours Newbie friendliness 45/100

llvm/torch-mlir#4744 · Opened Sep 2, 2026
[TorchToLinalg] Dynamic batch + sequence torch.nn.MultiheadAttention produces silently incorrect output; Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

llvm/torch-mlir#4743 · Opened Sep 1, 2026
[RFC] Use `core_aten_decompositions()` as the FX-path decomposition table Open

Difficulty 5/5 Over a week Newbie friendliness 42/100

llvm/torch-mlir#4729 · 5 comments · Opened Aug 20, 2026
[onnx] Support com.microsoft::Attention contrib op (onnxruntime-genai model builder) Open

Difficulty 4/5 3-5 days Newbie friendliness 68/100

llvm/torch-mlir#4726 · Opened Aug 20, 2026
[TorchToLinalg] Segfault instead of a clean failure when lowering 8-bit integer matmul / mm / bmm / convolution Open

Difficulty 4/5 3-5 days Newbie friendliness 68/100

llvm/torch-mlir#4725 · Opened Aug 20, 2026
[TorchToLinalg] Unconventional lowering of KernelBench softmax kernel Open

llvm/torch-mlir#4700 · 4 comments · 1 assignee · Opened Jul 30, 2026
[TorchOnnxToTorch] Generate unexpected unflatten for ScatterND Open

Difficulty 3/5 1-2 days Newbie friendliness 68/100

llvm/torch-mlir#4699 · 1 comment · Opened Jul 30, 2026
[TorchToLinalg] Missing lowering for torch.aten._scaled_mm Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

llvm/torch-mlir#4693 · Opened Jul 28, 2026
Torch-to-Linalg lowering fails to legalize torch.vtensor.literal for Hugging Face TinyLlama forward export Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

llvm/torch-mlir#4625 · 1 comment · Opened Jun 29, 2026
[TorchToLinalg] torch-mlir fails to legalize torch.aten.empty_permuted Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

llvm/torch-mlir#4619 · 1 comment · Opened Jun 22, 2026
[TorchToLinalg] torch-mlir fails to legalize torch.transformers.grouped_mm_fallback Open

Difficulty 4/5 3-5 days Newbie friendliness 55/100

llvm/torch-mlir#4618 · 1 comment · Opened Jun 22, 2026
[TorchToLinalg] torch-mlir fails to legalize torch.aten.histc Open

Difficulty 3/5 1-2 days Newbie friendliness 55/100

llvm/torch-mlir#4617 · Opened Jun 22, 2026
Inconsistent lowering of 3D and 4D matmuls Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

llvm/torch-mlir#4610 · 4 comments · Opened Jun 16, 2026
[RFC] First-class PT2E quantization support in torch-mlir Open

Difficulty 5/5 Over a week Newbie friendliness 35/100

llvm/torch-mlir#4600 · 20 comments · 4 reactions · Opened Jun 11, 2026
[Feature Request]: Add macOS wheels (arm64 / universal2) for torch-mlir dev releases Open

Difficulty 4/5 3-5 days Newbie friendliness 52/100

llvm/torch-mlir#4599 · 3 comments · Opened Jun 10, 2026
error: failed to legalize operation 'torch.operator' that was explicitly marked illegal Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

llvm/torch-mlir#4575 · 3 comments · Opened May 19, 2026
Dialect `tm_tensor` not found for custom op 'tm_tensor.scan' Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

llvm/torch-mlir#4574 · 5 comments · Opened May 19, 2026
onnx=1.16.1 requirement install from test-requirements.txt fails. Open

llvm/torch-mlir#4555 · 1 comment · 1 assignee · Opened May 4, 2026
[TorchONNXToTorch] Failed to lower torch.operator "onnx.Einsum" Open

Difficulty 4/5 3-5 days Newbie friendliness 48/100

llvm/torch-mlir#4535 · 1 comment · Opened Apr 14, 2026
How to pass in second argument when using torch.fx.export_and_import if forward function takes in multiple inputs. Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

llvm/torch-mlir#4524 · Opened Apr 6, 2026
[Feature Request]: Provide a clear overview or query method for supported / lowered Torch Aten Ops coverage in torch-mlir Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4503 · 1 comment · 1 reaction · Opened Mar 15, 2026
[TorchToLinalg] Failed to legalize operation 'torch.aten.index.Tensor_hacked_twin'. Open

llvm/torch-mlir#4495 · 3 comments · 1 assignee · Opened Mar 10, 2026
`DecomposeAtenNativeBatchNormOp` uses wrong dtype for running stats reshape Open

llvm/torch-mlir#4480 · 1 assignee · Opened Mar 2, 2026
`aten.view` fold crashes on element type mismatch instead of verifying Open

llvm/torch-mlir#4479 · 1 assignee · Opened Mar 2, 2026
Add support for aten.miopen_batch_norm Open

llvm/torch-mlir#4476 · 1 comment · 1 assignee · Opened Feb 23, 2026
Compile pytorch complexYOLOv3 model to MLIR failed Open

Difficulty 4/5 3-5 days Newbie friendliness 25/100

llvm/torch-mlir#4451 · Opened Jan 28, 2026
Torch-MLIR build takes up over 63 gigabytes Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4450 · Opened Jan 28, 2026
torch.export failed while exporting alpamayo model (Qwen3_vl) with symbolic error Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4436 · 2 comments · Opened Jan 21, 2026
`torch.aten.mm` fails to legalize with mixed precision operands Open

llvm/torch-mlir#4422 · 1 assignee · Opened Jan 5, 2026
Crash in GlobalSlotModuleInitializerOp verifier due to unexpected terminator type Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

llvm/torch-mlir#4414 · Opened Dec 22, 2025
Assertion failure in Block::getTerminator during verification of GlobalSlotModuleInitializerOp Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

llvm/torch-mlir#4413 · Opened Dec 22, 2025
Assertion failure in ValueTensorType::get during type inference from signless integer attributes Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

llvm/torch-mlir#4412 · Opened Dec 22, 2025
Crash in GlobalSlotModuleInitializerOp verifier due to invalid attribute type in slotSymNames Open

Difficulty 3/5 1-2 days Newbie friendliness 48/100

llvm/torch-mlir#4411 · Opened Dec 22, 2025
Assertion failure in SymbolTable triggered by duplicate torch.global_slot symbol names Open

Difficulty 3/5 1-2 days Newbie friendliness 48/100

llvm/torch-mlir#4410 · Opened Dec 22, 2025
[AdaptiveAvgPool2dOp] why the operator need 2 similar version ？ Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

llvm/torch-mlir#4393 · Opened Dec 2, 2025
ERROR: missing input file '@@+_repo_rules3+torch-mlir//:include/torch-mlir-dialects/Dialect/TMTensor/Transforms/PassDetail.h Open

Difficulty 3/5 1-2 days Newbie friendliness 35/100

llvm/torch-mlir#4392 · 1 comment · Opened Dec 2, 2025
how to compile scaled_dot_product_attention to iree--.vmfb? Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4379 · 1 comment · Opened Nov 17, 2025
[torch.logit] how to use one input version interface ? Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4374 · Opened Nov 12, 2025
[API] Strange interface name aten.conv_transpose3d.input Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4361 · Opened Nov 1, 2025
QLinearAdd operation fails to legalize during IREE compilation Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4357 · 1 comment · Opened Oct 28, 2025
Improve lowering of backward convolution to linalg Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4355 · 1 reaction · Opened Oct 24, 2025
In-Place PyTorch Ops Lower Incorrectly to Out-of-Place Open

Difficulty 4/5 3-5 days Newbie friendliness 38/100

llvm/torch-mlir#4350 · 1 comment · Opened Oct 17, 2025
error: 'torch.aten.copy_' op operand #0 must be Multi-dimensional array modeling Torch's Tensor type, but got '!torch.vtensor Open

Difficulty 4/5 3-5 days Newbie friendliness 35/100

llvm/torch-mlir#4346 · 1 comment · Opened Oct 15, 2025
error: POWI exponent does not match sizeof(int) Open

Difficulty 3/5 1-2 days Newbie friendliness 45/100

llvm/torch-mlir#4336 · Opened Oct 8, 2025
A new case not handled by aten.cat length-0 canonicalization Open

Difficulty 3/5 1-2 days Newbie friendliness 42/100

llvm/torch-mlir#4333 · 1 comment · Opened Oct 7, 2025
Dynamic shape support in TOSA backend? Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4321 · 6 comments · Opened Sep 22, 2025
Decomposing `aten.as_strided` causes assertion failure in MLIR Open

llvm/torch-mlir#4315 · 2 comments · 1 assignee · Opened Sep 17, 2025
[fx-importer] the importer is slow Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4312 · Opened Sep 15, 2025
matmulOp miss the check with the dim of result Open

Difficulty 3/5 1-2 days Newbie friendliness 52/100

llvm/torch-mlir#4309 · Opened Sep 9, 2025
Maxunpool does not support overlapping Maxpool inputs Open

Difficulty 5/5 Over a week Newbie friendliness 25/100

llvm/torch-mlir#4308 · 1 comment · Opened Sep 7, 2025

llvm/torch-mlir

Get new issues in your inbox