strisunshinewentingwang/pytorch

Pull Requests Commits

[Dynamo] Fix bugs when calling tensor.data and tensor.layout (#89257)

yanboliang committed 3 years ago

186192bb

[4/n] Thread PG: add reduce_scatter to threaded pg (#89442)

wanchaol committed 3 years ago

821ba6b5

[3/n] Thread PG: add scatter to threaded pg (#89441)

wanchaol committed 3 years ago

3e99d4db

[2/n] Thread PG: add test for broadcast (#89440)

wanchaol committed 3 years ago

3876f94c

[1/n] Thread PG: add test for allgather (#89439)

wanchaol committed 3 years ago

deae4508

[tools] expose selective build library (#89351)

larryliu0820 committed 3 years ago

047e542a

[inductor] Misc division lowering fixes (#88603)

peterbell10 committed 3 years ago

c068fa90

[inductor] Fix nan handling for aten.sign (#88937)

peterbell10 committed 3 years ago

1267dcf2

Fix unconvertible_ops as per #89261 (#89299)

kevalmorabia97 committed 3 years ago

3d247a8b

Update sdp dispatch logic to enable fused backward (#89154)

drisspg committed 3 years ago

1d9e1fca

update kineto pinned commit (#89435)

Taylor Robie committed 3 years ago

cf947655

Port torchdynamo's torchbench script to userbenchmark (#89239)

xuzhao9 committed 3 years ago

e4d9dbd7

Revert "[ao] making _is_activation_post_process private (#87520)"

pytorchmergebot committed 3 years ago

9d209e78

Revert "[ao] maintain BC for is_activation_post_process (#89260)"

pytorchmergebot committed 3 years ago

f3db0361

[Inductor] Limit the number of compile threads to the available cpu cores (#89377)

Jiong Gong committed 3 years ago

6796979e

Move the OpInfo same-storage error to the autograd test (#88306)

lezcano committed 3 years ago

c2cf0bde

Update ideep for future performance improvement (#87966)

yanbing-j committed 3 years ago

a80e5e78

TorchDynamo: enable conv+silu fusion (#89278)

XiaobingSuper committed 3 years ago

31708a73

Redefine the simdlen semantic (#89263)

EikanWang committed 3 years ago

bc716383

TorchDynamo: enable conv+relu6 fusion (#89265)

XiaobingSuper committed 3 years ago

79770d36

[Easy] Use prepend arg to register forward hooks in quantize.py (#89391)

mrshenli committed 3 years ago

e0251de4

[vision hash update] update the pinned vision hash (#89287)

pytorchmergebot committed 3 years ago

1db5ce09

use std/libdevice erf in inductor (#89388)

Natalia Gimelshein committed 3 years ago

51e961dd

Temporary increase ASAN shard 5 to 4xlarge (#89387)

huydhn committed 3 years ago

1856fa5d

Revert "Update sdp dispatch logic to enable fused backward (#89154)"

pytorchmergebot committed 3 years ago

e1d58b19

Also include MKL_THREAD_LIB in link libraries for caffe2::mkl (#89378)

ezyang committed 3 years ago

c0992965

Set INTERFACE_LINK_DIRECTORIES on caffe2::mkl (#89359)

ezyang committed 3 years ago

7b0d577c

Fix cat striding in PrimTorch (#89332)

ezyang committed 3 years ago

dbeacf11

Symintify numel(), infer_size, prims.elementwise_meta (#88956)

SherlockNoMad committed 3 years ago

caf3d531

Add support for dynamic kwarg to torch._dynamo.optimize (#89290)

ezyang committed 3 years ago

7c811efa

Newer Older