strisunshinewentingwang/pytorch

Pull Requests Commits

rohan-varma committed 2 years ago

08eb963d

rohan-varma committed 2 years ago

514b0809

[lint] preoperly suggest lintrunner when lint fails

suo committed 2 years ago

6a44e066

Revert "Add ZT fastpath for remainder.Tensor JVP"

pytorchmergebot committed 2 years ago

da15d76e

[complex32] real and imag (also remove unused real and imag kernels)

kshitij12345 committed 2 years ago

0f1618ef

[PT-D][Sharding] Clean up sharded tensor code by leverage handle_torch_function (#76824)

fduwjj committed 2 years ago

4c504dd3

Add ZT fastpath for remainder.Tensor JVP

Nikita Karetnikov committed 2 years ago

7471f614

Allow creating new libraries and defining new operators from Python

anjali411 committed 2 years ago

07f766df

Allow users to override kernels for existing C++ ops through Python

anjali411 committed 2 years ago

55f55a4c

ci: Add upload credentials for rocm periodic test

seemethere committed 2 years ago

32ae5840

[ONNX] Format ONNX python with black

justinchuby committed 2 years ago

5dd1c677

Return all overloads for an operator in _jit_get_operation

ezyang committed 2 years ago

3a6da16a

Change TensorMeta to use tensor subclass.

ezyang committed 2 years ago

4a116783

Use TorchFunctionMode to implement PrimTorch tracing context

ezyang committed 2 years ago

48eb8d6a

[LT] Add a trie data structure for caching IR nodes

desertfire committed 2 years ago

f05710dd

Dispatch to mv rather than mm in the case that tensor1.ndim == 1 and tensor2.ndim == 2

lezcano committed 2 years ago

4baf7c08

Fix mv/addmv on CUDA when dealing with vectors of size=1 and stride=0

lezcano committed 2 years ago

02b5b92c

Refactor the API of the matmul implementation

lezcano committed 2 years ago

35ae9f68

[SR] Eliminate extra permutes around softmax calls (#76391)

Mike Iovine committed 2 years ago

1fed6b75

[SR] Codegen for aten::clamp (#76340)

Mike Iovine committed 2 years ago

cac2733a

Remove pow and float_power TestGradient Skips

drisspg committed 2 years ago

c59d5f17

Revert "Use scatter_reduce to support masked reductions on sparse COO tensors (sum, prod, amin, amax)"

pytorchmergebot committed 2 years ago

381e0830

[complex32] enable testing for multiple ops

kshitij12345 committed 2 years ago

8a2f207d

Sparse CSR: Add CPU fallback for sampled_addmm

IvanYashchuk committed 2 years ago

13355120

Added logit/reciprocal decomps, fixed var for complex, moved type promotion logic to standardize on primtorch's

Chillee committed 2 years ago

6917034a

Revert "Bugfix NAN and Inf handling for scatter_reduce (amin and amax)"

pytorchmergebot committed 2 years ago

4aa6b6b9

Fix unmarked fstring

peterbell10 committed 2 years ago

6df5a531

DOC fix momentum equation for nesterov

tomMoral committed 2 years ago

ff94c9de

Add high level control of fp32 matmul precision; disable TF32 for matmuls by default

eqy committed 2 years ago

e838137b

[ONNX] Support optional type (#68793) (#73284)

BowenBao committed 2 years ago

679fc90c

Older