strisunshinewentingwang/pytorch

Pull Requests Commits

Revert [DDP] Support not all outputs used in loss calculation (#61497)

rohan-varma committed 3 years ago

71775093

Revert [DDP] Support for multiple backwards (#61401)

rohan-varma committed 3 years ago

25f9c35d

Back out "[DDP] Disable reducer hooks from running outside of DDP backwards." (#61399)

rohan-varma committed 3 years ago

38ac9e69

Revert D29701479: [pytorch][PR] Remove `_broadcast_object()` from `ZeroRedundancyOptimizer`

yuguo68 committed 3 years ago

a50a389c

Fix for get_buffer(): check buffers by name instead of value (#61429)

jbschlosser committed 3 years ago

aa01a7a6

CopyBackward: Remove redundant src_device and unnecessary copy=True (#60025)

peterbell10 committed 3 years ago

54071085

Add .github for CODEOWNERS (#61598)

seemethere committed 3 years ago

da667e2d

changed launch bounds for upsample_linear1d fwd, bwd from 1024 to 512 (#61307)

Fuzzkatt committed 3 years ago

8afb65b6

Register Saved Tensors hooks (#60663)

Varal7 committed 3 years ago

ee5a97de

[static runtime] Use at::allclose to test NNC sigmoid (#61566)

d1jang committed 3 years ago

94965212

Remove `_broadcast_object()` from `ZeroRedundancyOptimizer` (#61539)

awgu committed 3 years ago

9b5d9b40

[pytorch][profiler] Fix division by 0 in computeFlops (#61676)

valentinandrei committed 3 years ago

e3d5619f

Avoid redefining `__BYTE_ORDER` (#60346)

Dimitrije Jankov committed 3 years ago

70e94bb1

Grammatical update of tech docs (#61547)

JINAY08 committed 3 years ago

a9c35800

add trainer hook functions (#60785)

gcramer23 committed 3 years ago

5a5c7f56

refactor ps benchmark (#60784)

gcramer23 committed 3 years ago

304c02ee

Release GIL as much as possible in dist_autograd pybind. (#61593)

pritamdamania committed 3 years ago

7d2ea9a8

Avoid holding GIL while calling retrieveContext. (#61588)

pritamdamania committed 3 years ago

5ebc7c9f

[Metal] Do not use read/write textures in concat shaders (#61074)

SS-JIA committed 3 years ago

f2adbff3

Skip Bfloat16 support when building for VSX (#61630)

malfet committed 3 years ago

80bdfd64

[TensorExpr] Do not fuse float16 values. (#61569)

Mikhail Zolotukhin committed 3 years ago

43a2f7c2

Make broadcast_object_list accept a device parameter. (#61305)

bowangbj committed 3 years ago

ab273995

[pruner] fix activation handles logic (#61592)

kazhou committed 3 years ago

9b3cbeaf

[pytorch] Add broadcasting support to add_relu kernel (#61584)

John Shen committed 3 years ago

343cb276

Smart Decay for Adam - Caffe2 (#61548)

Jamie King committed 3 years ago

c23db932

Enable C2 load rate limiter [2/n] (#61551)

Kaige Liu committed 3 years ago

58adaaba

Refactor non-joined process computation (#61555)

awgu committed 3 years ago

57feb354

adding support for index_select on quantized tensors (#61406)

HDCharles committed 3 years ago

03a79f43

[Static Runtime] Check unsupported up when enabling static runtime (#61613)

Hao Lu committed 3 years ago

a07b0813

[FX][docs] Add note about python set pitfall (#61597)

James Reed committed 3 years ago

ac64a41e

Newer Older