llvm/torch-mlir
The Torch-MLIR project aims to provide first class support from the PyTorch ecosystem to the MLIR ecosystem.
C++NOASSERTION
Issues
- 5
- 6
[ONNX] Add support for onnx.Resize with nearest_mode "round_prefer_floor" and "round_prefer_ceil"
#3420 opened by flemairen6 - 0
[Query] The refback-munge-calling-conventions pass employed in the backend from linalg to llvm, causes runtime error
#3682 opened by Abhishek-TyRnT - 0
[Linalg] linalg.matmul f16 type convert to f32
#3662 opened by CoTinker - 3
- 1
convert-torch-onnx-to-torch generates invalid IR for onnx.Resize where scaling is in the first two dimensions
#3453 opened by mgehre-amd - 3
Failed to convert "onnx.Pad" into torch
#3442 opened by josel-amd - 1
- 4
Issue with lowering GPT2 to linalgIR
#3649 opened by sdalvi-quic - 4
- 0
Empty `torch_mlir._mlir_libs._jit_ir_importer` lib causes package build problem on macOS
#3663 opened by dbabokin - 1
Error compiling MobilenetV2
#3533 opened by marcospazianibrunella - 6
Missing dimensionality information in torch
#3651 opened by MaheshRavishankar - 8
torch_mlir.tools.import_onnx misses shape inference
#3593 opened by jinchen62 - 0
aten.Multinomial tries to add f32 and f64 together
#3628 opened by zjgarvey - 0
[FXImporter] Find a better way to get a buffer representation of a torch.Tensor (support bfloat and complex dtypes when faced with lift_fresh_copy)
#3653 opened by 123epsilon - 0
There is an incorrect hyperlink in docs/ltc_examples.md
#3440 opened by simonzgx - 0
torch.aten.remainder.Tensor op produces incorrect results when calculating the remainder between a tensor with negative values and a tensor with positive values
#3575 opened by BaneTrifa - 2
- 1
- 0
- 1
When compile custom_op, got error: failed to legalize operation 'torch.operator' that was explicitly marked illegal
#3596 opened by 0x00-pl - 0
- 0
- 1
- 3
- 4
One or more Stablehlo test(s) crashing after llvm bump to 266a5a9cb9daa96c1eeaebc18e10f5a37d638734
#3549 opened by aviator19941 - 7
Need E2E ONNX op tests in CI
#3520 opened by renxida - 0
Reduction ops don't support dynamic dims
#3573 opened by renxida - 0
- 0
- 0
Dtype checks failing with 'Subtraction, the `-` operator, with two bool tensors is not supported. Use the `^` or `logical_xor()` operator instead.'
#3552 opened by vivekkhandelwal1 - 1
- 1
Can't lower onnx.Pad with "wrap" mode
#3532 opened by josel-amd - 1
- 0
[ONNX] Bernoulli operator implementation might be wrong (value mismatch in e2e testing when using function expansion)
#3527 opened by andfau-amd - 2
- 2
Bump onnx to 1.16.1
#3507 opened by stellaraccident - 4
support torch.aten.diag
#3499 opened by aartbik - 0
MLIR Python pass manager API will assert if the pass issues errors but does not signal failure
#3506 opened by stellaraccident - 3
- 7
- 0
- 2
Error lowering `index_put_`: 'tm_tensor.scatter' op mismatch in shape of indices and update value at dim#0
#3433 opened by ScottTodd - 0
Remove optional tensor logic from linalg lowering for `IndexPutHackedTwinOp`
#3489 opened by ramiro050 - 0
- 3
- 6
Failed to lower "torch.aten.sum.dim_IntList" to linalg
#3423 opened by josel-amd - 3
Failed to lower "torch.aten.convolution" to linalg
#3429 opened by josel-amd - 0
Failed to lower batch normalization (in training mode)
#3438 opened by josel-amd