main

GGitHub[Torch] Canonicalize pool ops with single int tuple params. (#4250)

244f4b60创建于 2025年11月24日历史提交

文件	最后提交记录	最后更新时间
jit_ir_common	Fix sanitizers issues (#4050) ubsan detected 'lib/Dialect/TorchConversion/Transforms/UnpackQuantTensor.cpp:109:21: runtime error: left shift of negative value -16' and asan detected memcpy with nullptr when we passed and empty vector with .data() == nullptr to mlirOperationStateAddResults. Detected using https://github.com/llvm/torch-mlir/pull/4046	1 年前
ltc	Add shape inference for AtenAsStrided (#4076) Added shape inference for AtenAsStridedOp	1 年前
onnx_c_importer	Bring the C++ ONNX importer on par with onnx_importer.py (#3960) This PR heavily refactors the C++ importer to make it a viable (and faster) alternative to the Python one for those who need to import ONNX models and want to avoid depending on the python ecosystem. - The C++ importer now outputs the same exact mlir as the Python one (tested on alt_e2eshark test suite). Achieving perfect output matches required to introduce an associative map iterable according to insertion order (to mimic Dict in Python). - The code tries to mirror 1-to-1 the Python counterpart whenever possible/convenient. - Adds support for embedding ONNX external data in the mlir. This functionality is not part of torch-mlir's onnx_importer.py but of IREE's import_onnx. - Efforts have been made to remove the direct dependency on LLVM support lib. There is however a transitive dependency on such lib through MLIRCAPIIR and TorchMLIRCAPI (MLIR libraries uniformly depend on LLVMSupport).	1 年前
pt1	[Torch] Canonicalize pool ops with single int tuple params. (#4250) Fixes https://github.com/llvm/torch-mlir/issues/3885 by repeating the single int to match with expected spatial dims.	8 个月前
CMakeLists.txt	[NFC reformat] Run pre-commit on all files and format misc. This is part 1 of ~3, formatting all miscellaneous text files and CPP files matched by a first run of pre-commit. These tend to be low change-traffic and are likely not disruptive. Subsequent patches will format Python files and remaining CPP files.	2 年前