strisunshinewentingwang/pytorch

Pull Requests Commits

[FX] Fix GraphModule copy methods not regenerating forward (#44806)

James Reed committed 4 years ago

60ae6c9c

[reland] move rebuild buckets from end of first iteration to beginning of second iteration (#44798)

zhaojuanmao committed 4 years ago

e14b2080

Enable torch.backends.cuda typechecking in CI (#44916)

malfet committed 4 years ago

2043fbdf

[TensorExpr] Add Mod support to the LLVM backend (#44823)

asuhan committed 4 years ago

18b77d7d

[ONNX] Enable true_divide scripting export with ONNX shape inference (#43991)

BowenBao committed 4 years ago

e535fb3f

Enable typechecking for torch.testing._internal.common_quantized.* (#44805)

janeyx99 committed 4 years ago

1c996b71

[TensorExpr] Fix order comparisons for unsigned types (#44857)

asuhan committed 4 years ago

f5b92332

Let logspace support bfloat16 on both CPU and CUDA (#44675)

xuhdev committed 4 years ago

a153eafa

Make nuclear and frobenius norm non-out depend on out variants (#44095)

kurtamohler committed 4 years ago

40e44c5f

[caffe2] add cost inference for FusedFakeQuantFC and FusedFakeQuantFCGradient (#44840)

jspark1105 committed 4 years ago

086a2e7a

Do not use `PRId64` in torch/csrc (#44767)

malfet committed 4 years ago

40660221

[TensorExpr] Add log1p support to the LLVM backend (#44839)

asuhan committed 4 years ago

5d570252

CUDA BFloat16 i0 support (#44750)

zasdfgbnm committed 4 years ago

f5440a44

Document the default behavior for dist.new_group() when ranks=None (#44000)

rohan-varma committed 4 years ago

bee97d5b

Implement sort for list of tuples (#43448)

gmagogsfm committed 4 years ago

2558e576

CUDA BFloat16 unary ops part 2 (#44824)

zasdfgbnm committed 4 years ago

c189328e

fix legacy GET_BLOCKS code from THCUNN/common.h (#44789)

Rong Rong committed 4 years ago

c1fa4249

torch.empty_like and torch.zeros_like raise error if any memory format is provided with sparse input (#43699) (#44058)

vfdev-5 committed 4 years ago

24df3b73

[quant][qat] Ensure fake_quant and observer can be disabled on scriptmodule (#44773)

supriyar committed 4 years ago

1fde54d5

[quant][fx] Add node name as prefix to observer module name (#44765)

supriyar committed 4 years ago

361b38da

Revert D23725053: [pytorch][PR] change self.generator to generator

ngimel committed 4 years ago

74c3dcd1

refactor intialize bucket views (#44330)

zhaojuanmao committed 4 years ago

d2b4534d

.circleci: Switch to dynamic MAX_JOBS (#44729)

seemethere committed 4 years ago

6006e450

Implement better caching allocator for segmentation usecase. (#44618)

kimishpatel committed 4 years ago

f605d758

minor style edits to torch/testing/_internal/common_quantized.py (#44807)

janeyx99 committed 4 years ago

4affbbd9

[te] Disable flaky test CudaSharedMemReduce_1 (#44862)

bertmaher committed 4 years ago

a40ef25e

Always use NewModuleTest instead of ModuleTest. (#44745)

gchanan committed 4 years ago

503c7488

Fixed quantile nan propagation and implemented nanquantile (#44393)

heitorschueroff committed 4 years ago

28085cbd

Support Python Slice class in TorchScript (#44335)

gmagogsfm committed 4 years ago

99093277

Revert D23236088: [pytorch][PR] [caffe2] adds Cancel to SafeDequeueBlobsOp and SafeEnqueueBlobsOp

Mike Ruberry committed 4 years ago

b6f4bb0a

Newer Older