strisunshinewentingwang/pytorch

Pull Requests Commits

Add missing file to BUILD.bazel (#40536)

malfet committed 4 years ago

e12f73ee

Use tree-based sum for floats to avoid numerical instability (#39516)

peterbell10 committed 4 years ago

3dcc3297

Release GIL during DDP construction. (#40495)

pritamdamania committed 4 years ago

ea06db94

Update FP16 to FP16:4dfe081cf6bcd15db339cf2680b9281b8451eeb3. (#40526)

Ashkan Aliabadi committed 4 years ago

71edd7f1

Add C++-only `int dim` overloads to `std`-related operations (#40451)

peterbell10 committed 4 years ago

16f276ce

Update cpuinfo to cpuinfo:63b254577ed77a8004a9be6ac707f3dccc4e1fd9. (#40516)

Ashkan Aliabadi committed 4 years ago

a208a272

Unify `torch/csrc/cuda/shared/cudnn.cpp` include path (#40525)

malfet committed 4 years ago

c120fdc0

Update FXdiv to FXdiv:b408327ac2a15ec3e43352421954f5b1967701d1. (#40520)

Ashkan Aliabadi committed 4 years ago

cef35e33

Update psimd to psimd:072586a71b55b7f8c584153d223e95687148a900. (#40522)

Ashkan Aliabadi committed 4 years ago

4a0ba62d

[jit] allow dict to be mixed between tracing and scripting (#39601)

wanchaol committed 4 years ago

3e09268c

[jit] fix dictConstruct order issue (#40424)

wanchaol committed 4 years ago

787e1c4c

Update docs feature classifications (#39966)

Jessica Lin committed 4 years ago

2e6e8d55

Migrate equal from the TH to Aten (CPU) (#33286)

xcnick committed 4 years ago

72f2c479

Skip test_mem_leak on Windows (#40486)

peterjc123 committed 4 years ago

4d549077

Add finishAndThrow function to ProcessGroup::Work, and use with Gloo (#40405)

osalpekar committed 4 years ago

0c923eea

[NCCL Docs] Adding Comments for Work-level Finish in ProcessGroup (#40404)

osalpekar committed 4 years ago

3e2d2fc8

[NCCL] Explicitly Abort NCCL Communicators on Process Group Destruction (#40241)

osalpekar committed 4 years ago

527ab134

Use GLOG logging prefixes (#40491)

malfet committed 4 years ago

fe18dcd6

enable mkldnn dilation conv (#40483)

XiaobingSuper committed 4 years ago

fc4824aa

Add out= variants for cuda.comm.broadcast/gather/scatter (#39681)

ssnl committed 4 years ago

de7ac60c

Adds dynamic versioning pattern (#40279)

Mike Ruberry committed 4 years ago

e6644587

Increase number of iterations in DDP SPMD tests (#40506)

mrshenli committed 4 years ago

a2e1a948

Add guard for non-default stream in DDP's autograd engine callback (#40115)

rohan-varma committed 4 years ago

9a3e16c7

Use Int8QuantParamsBlob to pass the scale and zeropoint params (#40494)

csummersea committed 4 years ago

597cb04b

Report error when ATEN_THEADING is OMP and USE_OPENMP is turned off. (#40146)

xuhdev committed 4 years ago

3ed96e46

Allow torch.cuda.amp.GradScaler to support sparse gradients (#36786)

definitelynotmcarilli committed 4 years ago

b4ccdef0

wconstab/38034-sliced-sequential (#40445)

wconstab committed 4 years ago

d8555281

Initial vmap frontend API (#40172)

zou3519 committed 4 years ago

727463a7

Add invariants check to BatchedTensorImpl (#40171)

zou3519 committed 4 years ago

43ab9c67

Simplify complex case for tanh backward (#39997)

ShawnZhong committed 4 years ago

e490352d

Newer Older