strisunshinewentingwang/pytorch

Pull Requests Commits

check error status of CUDA launch after Magma kernels (#29003)

soumith committed 4 years ago

f441bb1c

Revert "Revert D18171156: Merge Tensor and Variable." (#29299)

ezyang committed 4 years ago

4e21157e

Add functional overloads for activation, batchnorm, distance, embedding (#29358)

Will Feng committed 4 years ago

b24b967e

Revert RRef.to_here()/local_value() return type (#29396)

mrshenli committed 4 years ago

63675b19

Dump operator names of a module and its submodules.

Martin Yuan committed 4 years ago

d75222f3

Clean up the stale item in bc white list (#29439)

houseroad committed 4 years ago

b7fc26a9

don't copy ignored/unused methods to ScriptModule (#29342)

suo committed 4 years ago

255b2340

Add note to docs of torch.unique (#29165)

zasdfgbnm committed 4 years ago

5f03ad96

Skips CUDA handle tests on Python2 (#29430)

Mike Ruberry committed 4 years ago

baef925d

Make HistogramObserver scriptable with `@torch.jit.ignore` (#27950)

jerryzh168 committed 4 years ago

4bcf4796

fix negative string indexing (#22700)

Elias Ellison committed 4 years ago

19d3a7ad

Lift rpc_timeout to RpcAgent, for other RpcAgents to reuse. (#29341)

xush6528 committed 4 years ago

e66626ae

Export weight_norm (#28618)

neginraoof committed 4 years ago

7da11f49

Make jit.trace_module reentrant (#29411)

James Reed committed 4 years ago

782e80e6

enable fast path for TensorIterator for contiguous inputs/no broadcast (#29180)

ngimel committed 4 years ago

90f28c27

Use nativeloader instead of system loader to load JNI library for soloader compatibility. (#29350)

xcheng16 committed 4 years ago

8a33f115

Simplify _calculate_fan_in_and_fan_out (#29370)

nzmora committed 4 years ago

fa66a149

clone should preserve the type of attribute (#29269)

jerryzh168 committed 4 years ago

de9a5446

fix pytorch mobile build (#29414)

linbinyu committed 4 years ago

5a441071

Updating submodules

svcscm committed 4 years ago

0be2f12e

Fix tracing for dynamic quantized LSTM (#29331)

James Reed committed 4 years ago

821f8bfc

Updating submodules

svcscm committed 4 years ago

5bb35fe9

Skip flaky test

James Reed committed 4 years ago

1dd3c8e5

Use cuDNN's handle pool mechanism to manage cublas handles (#29233)

zasdfgbnm committed 4 years ago

02921e79

explicitly provide memory format when calling to clone() at SparseTensor.cpp

ifedan committed 4 years ago

b008b34b

explicitly provide memory format when calling to clone() at SparseTensorMath.cpp

ifedan committed 4 years ago

09822a1d

Automatic update of fbcode/onnx to fea8568cac61a482ed208748fdc0e1a8e47f62f5 (#29363)

houseroad committed 4 years ago

564384fe

Updating submodules

svcscm committed 4 years ago

255505f2

explicitly provide memory format when calling to clone() at TensorShape.cpp

ifedan committed 4 years ago

d5d524da

NHWC support in cuDNN BatchNorm & Conv2d (#29361)

jjsjann123 committed 4 years ago

fdab1cf0

Newer Older