pytorch

30675d09 - Added OpInfo-based testing of triangular_solve (#50948)

Added OpInfo-based testing of triangular_solve (#50948) Summary: Added OpInfo-based testing of `torch.triangular_solve`. These tests helped to discover that CPU `triangular_solve` wasn't working for empty matrices and for CUDA inputs a warning was printed to the terminal. It is fixed now. CUDA gradgrad checks are skipped. ``` 11.44s call test/test_ops.py::TestGradientsCUDA::test_fn_gradgrad_triangular_solve_cuda_complex128 2.97s call test/test_ops.py::TestGradientsCUDA::test_fn_gradgrad_triangular_solve_cuda_float64 1.60s call test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_triangular_solve_cpu_complex128 1.36s call test/test_ops.py::TestOpInfoCUDA::test_supported_dtypes_triangular_solve_cuda_complex128 1.20s call test/test_ops.py::TestGradientsCUDA::test_fn_grad_triangular_solve_cuda_complex128 0.86s call test/test_ops.py::TestCommonCUDA::test_variant_consistency_jit_triangular_solve_cuda_complex64 0.85s call test/test_ops.py::TestCommonCUDA::test_variant_consistency_jit_triangular_solve_cuda_complex128 0.81s call test/test_ops.py::TestCommonCUDA::test_variant_consistency_jit_triangular_solve_cuda_float64 0.77s call test/test_ops.py::TestCommonCUDA::test_variant_consistency_jit_triangular_solve_cuda_float32 0.46s call test/test_ops.py::TestCommonCPU::test_variant_consistency_jit_triangular_solve_cpu_complex128 0.44s call test/test_ops.py::TestCommonCPU::test_variant_consistency_jit_triangular_solve_cpu_complex64 0.44s call test/test_ops.py::TestGradientsCUDA::test_fn_grad_triangular_solve_cuda_float64 0.42s call test/test_ops.py::TestGradientsCPU::test_fn_gradgrad_triangular_solve_cpu_float64 0.40s call test/test_ops.py::TestCommonCPU::test_variant_consistency_jit_triangular_solve_cpu_float32 0.40s call test/test_ops.py::TestCommonCPU::test_variant_consistency_jit_triangular_solve_cpu_float64 0.17s call test/test_ops.py::TestGradientsCPU::test_fn_grad_triangular_solve_cpu_complex128 ``` Ref. https://github.com/pytorch/pytorch/issues/50006 Pull Request resolved: https://github.com/pytorch/pytorch/pull/50948 Reviewed By: ailzhang Differential Revision: D26123998 Pulled By: mruberry fbshipit-source-id: 54136e8fc8a71f107dddb692c5be298c6d5ed168