strisunshinewentingwang/pytorch

Pull Requests Commits

[WIP] Layernorm Fake FP16 Op. (#39103)

venkatacrc committed 4 years ago

85b3fa03

More fixes about using Windows API through ctypes (#39376)

peterjc123 committed 4 years ago

30146d73

[TensorPipe] Acquire lock when adding message to timeout map (#39398)

lw committed 4 years ago

e358adb4

[TensorPipe] Guess IP addr in separate function (#39397)

lw committed 4 years ago

e142d703

Remove unboxed only from AMP registrations for cat. (#39156)

ezyang committed 4 years ago

de5b8797

The exception raised when RandomSampler.replacement is non-boolean should be TypeError (#36547)

xuhdev committed 4 years ago

283a3ff1

Clean up cast from c10::complex<T> to thrust::complex<T>, and update the workaround CUDA version to <10.2 (#38941)

xuhdev committed 4 years ago

413f0237

[pytorch] Let jit.unused ignore unsupported method signature (#39336)

ppwwyyxx committed 4 years ago

68f23d56

[JIT] Add support for saving/loading of lowered modules (#38893)

Meghan Lele committed 4 years ago

f4365cf5

ONNX Export Support for Celu (#38243)

KsenijaS committed 4 years ago

858ab750

Resubmit [Onnxifi] Generic way of passing output shape/type hints (#39377)

Yinghai Lu committed 4 years ago

ed26e8b0

[caffe2] Update shape info delimiter (#39275)

Hao Lu committed 4 years ago

f16b04f8

[quant][graphmode][fix] Run preprocess for child module before parent module (#39368)

jerryzh168 committed 4 years ago

f29fa06c

[quant] Fix fusion pattern for add_relu (#39367)

jerryzh168 committed 4 years ago

625f4e39

[doc] [distributed] fix typo (#39264)

ssnl committed 4 years ago

3001facd

[Futures] Allow setErrorIfNeeded arg to have type FutureError (#39113)

osalpekar committed 4 years ago

a47e2d44

Restore thrust path for 1d tensors cumulative ops (#39180)

Natalia Gimelshein committed 4 years ago

f1170898

Revert D21781515: [Onnxifi] Generic way of passing output shape/type hints

smessmer committed 4 years ago

e286cb5e

[Onnxifi] Generic way of passing output shape/type hints (#39229)

Yinghai Lu committed 4 years ago

7b07208d

Use @skipIfNoFBGEMM instead of direct check (#39068)

malfet committed 4 years ago

c5def603

Fix return types of Windows API functions in __init__.py (#39334)

peterjc123 committed 4 years ago

e6d86036

[Futures] Added markCompletedIfNeeded API (#39080)

osalpekar committed 4 years ago

295a23f4

Check illegal output dtype for torch.{min, max} (#38850)

zasdfgbnm committed 4 years ago

48e66859

Make Optimizer.state_dict() nondeterministic (#37347)

aphedges committed 4 years ago

a3c87c49

Adding sparse Lp regularization operator to Caffe2 (#38574)

Jamie King committed 4 years ago

7f1a96d4

Made sure torchscript compiles in optimized mode (#38888)

benoitsteiner committed 4 years ago

6d3e4aa0

Automated submodule update: FBGEMM (#39322)

facebook-github-bot committed 4 years ago

f76e05a2

Don't generate DeviceGuard for CPU wrapping code. (#38806)

ezyang committed 4 years ago

cffa0bee

Remove supports_named_tensor from codegen entirely. (#38739)

ezyang committed 4 years ago

2b6a48e9

[Profiler x RPC] Enable RPC Server Global Profiler (#38847)

xush6528 committed 4 years ago

45baf0e1

Newer Older