Yukio Siraichi
Yukio Siraichi
List of operations, whose `out=` variants are not consistent with eager (i.e. run with eager, but fail with dynamo). I have grouped them according to the error each of them...
Taking over: #7213 This PR introduces a test using PyTorch's `to_dlpack` and `from_dlpack` functions. In order to use the former, PyTorch's tensor `__dlpack__` and `__dlpack_device__` methods had to be modified....
This PR fixes #7998. Instead of always returning true, we forward this call to the base class `TensorImpl::is_contiguous_custom()`. The reason is that after https://github.com/pytorch/pytorch/pull/135498 is merged, XLA tensors' metadata might...
This PR aligns with https://github.com/pytorch/pytorch/pull/137525, which runs meta functions for `out=` variants. cc @JackCaoG
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #135498 * #135237 * __->__ #135497 This PR overrides more tensor metadata functions to the inner tensor inside `FunctionalTensorWrapper`. This is necessary so...
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #140287 Tracking issue: #138399 This PR changes the `pow` C++ implementation, making its C++ meta kernel consistent with its Python ref implementation....
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * #141078 * __->__ #141077 Fix: #139936 This PR modifies the lowering of `split` operation, so that it won't generate guards, specializing on the...
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #141078 * #141077 This PR turns clamping off for the `split` operation. By doing so, we generate less bound guards and reduce...
Stack from [ghstack](https://github.com/ezyang/ghstack) (oldest at bottom): * __->__ #135498 * #135237 * #135497 This PR modifies the `to_meta` function present in the functionalization layer, so that it also preserves the...