strisunshinewentingwang/pytorch

Pull Requests Commits

[vision hash update] update the pinned vision hash (#108611)

pytorchupdatebot committed 1 year ago

ce4967ad

Force synced KJT to trace unbacked SymInt (#107788)

ezyang committed 1 year ago

3b92ef81

Improve mem efficiency of constant folding (#108421)

eellison committed 1 year ago

c8e72a4a

[inductor] Use empty_strided to create output tensors when testing AOTInductor (#108364)

desertfire committed 1 year ago

28c5b622

[pytorch-vulkan] aten::.rand_like (#108086)

tina134 committed 1 year ago

d494b923

fix inline_container.cc inplace loading (#108573)

842974287 committed 1 year ago

d471eaeb

Fix dynamo benchmark config --print-graph-breaks (#108584)

yanboliang committed 1 year ago

ff28b4b9

Update clang7 CI jobs to clang9 (#108339)

cyyever committed 1 year ago

bae14b3d

[export] Fix dict.get() to dict.setdefault() for param lookup. (#108587)

zhxchen17 committed 1 year ago

c99a70c8

fix matrix_power documentation bug (#108585)

kjslag committed 1 year ago

eab57145

[RFC] Somewhat BC breaking: make checkpoint_wrapper default to NO_REENTRANT (#108435)

rohan-varma committed 1 year ago

208fd1cb

[RFC][FSDP] Don't move ignored params / buffers to device (#108033)

rohan-varma committed 1 year ago

db6d09c0

[RFC] Don't materialize ignored modules for FSDP (#108032)

rohan-varma committed 1 year ago

3334ec3a

[pytorch] Update docstring for FSDP.set_state_dict_type (#103864)

daniellepintz committed 1 year ago

fee9fc1d

[XNNPACK] Enable XX kernels (#108440)

digantdesai committed 1 year ago

64ad16a5

[HSDP] Add device_mesh to FSDP kwarg and add dtensor state_dict support for HSDP (#107533)

wz337 committed 1 year ago

66af4f6e

Fix doc preview page url at CONTRIBUTING.md (#108580)

Thiago Crepaldi committed 1 year ago

b1729d8b

fix issue with lift_fresh_copy when using export + compile (#108243)

bdhirsh committed 1 year ago

fac7a1f7

error when using _dynamo.optimize_ddp=True and _inductor.keep_output_stride=False together (#108235)

bdhirsh committed 1 year ago

da914aed

Fix inductor <> ddp_optimizer issue (#108081)

bdhirsh committed 1 year ago

def33d4d

Use global variables to register the return_types namedtuples (#107000)

andreigh committed 1 year ago

ae8eb7a3

Fix error message concatanation (#108581)

mthrok committed 1 year ago

d64e1c5f

[inductor] Update how AOTInductor resizes output tensors (#108412)

desertfire committed 1 year ago

7cdfc384

Revert "Use std::filesystem in c10 tempfile and tempdir (#106656)"

pytorchmergebot committed 1 year ago

1b76a5c2

Revert "[export] Copy gm before calling PassManager" for test or build failures (#108441)

digantdesai committed 1 year ago

a9a64232

fix use_deterministic_algorithms docstring (#108551)

kato8966 committed 1 year ago

0b44fdfa

[c10d] Introduce TCPStore client metrics collection. (#108348)

Rodrigo Kumpera committed 1 year ago

23e8a11f

[jit] Verify stack size and index to prevent off-by-one error (#108413)

apach301 committed 1 year ago

4a472d9e

torch.compile-functorch interaction: update docs (#108130)

kshitij12345 committed 1 year ago

a74f50d5

add Half support for maxpool on CPU (#98819)

CaoE committed 1 year ago

42f94d7e

Newer Older