strisunshinewentingwang/pytorch

Pull Requests Commits

Remove c++14-conditional constexpr (#30916)

smessmer committed 4 years ago

c21f8997

Fix IValue.isList

dreiss committed 4 years ago

4daa3ded

Properly return data from non-contiguous tensors in Java

dreiss committed 4 years ago

1b4d3d57

Support tensors with a storage offset in Java (#31584)

dreiss committed 4 years ago

2d6a2c89

Support tensors with empty shape in Java

dreiss committed 4 years ago

6d1fa829

Better error for `torch::jit::load`ing a eager file (#31709)

davidriazati committed 4 years ago

3c07eb33

Make RRef leak detection always print a warning log (#31922)

xush6528 committed 4 years ago

a730920a

Revert D18838848: disable __torch_function__ overides for operators in torch.functional

Karl Ostmo committed 4 years ago

227d1a43

Run a non-quiet submodule update to prevent timeouts on Circle CI (#31900)

Bram Wasti committed 4 years ago

8a0503b3

For torch::from_blob() add clue when memory is non-owned. (#31222)

jjlilley committed 4 years ago

114562cf

disable __torch_function__ overides for operators in torch.functional (#30839)

ngoldbaum committed 4 years ago

ca72df06

named tensor max pooling support

lixinyu committed 4 years ago

bb279c5c

Fix tracing for modules with List[Tensor] as output (#31343)

volkhin committed 4 years ago

3a2757c6

Raise an error if torch.cat is given `out` as one of the input tensors (#30577)

peterbell10 committed 4 years ago

74d69e29

Restructure docs organization and naming (#31849)

Jessica Lin committed 4 years ago

c888473b

Integrate async mode for autograd engine with distributed autograd. (#31508)

pritamdamania committed 4 years ago

bf8e1c07

Emit warning from deprecated torch function signatures (#31514)

peterbell10 committed 4 years ago

0e5a6700

Ensure autograd callbacks are called only once for reentrant backward. (#31909)

pritamdamania committed 4 years ago

5cc62f29

Support PyTorch ROCm CI on Ubuntu18.04 (#31886)

iotamudelta committed 4 years ago

4ee9c562

Raise ValueError if CUDA device is specified without specifying the : (#29087)

v0dro committed 4 years ago

2f5eefe5

Don't unconditionally compile runJITCPPTests (#31236)

ezyang committed 4 years ago

3c7db5cc

Enable personalized FC weight_init and sparse_emb weight_init (#31707)

Fei Tian committed 4 years ago

809ee9d0

Use TORCH_CHECK instead of AT_ASSERT in torch::cuda::gather() (#27456)

andreaskoepf committed 4 years ago

22044c6f

Add stub for transformer.py and MultiheadAttention Class. (#28396)

yyb1995 committed 4 years ago

20c5dd59

Update all instances of 1.4.0 -> 1.5.0 (#31785)

seemethere committed 4 years ago

346a3491

Enable BFloat16 support for Convolutions on ROCm (#30948)

rohithkrn committed 4 years ago

985fd970

minor doc tweak to use mp.spawn in example (#30381)

rohan-varma committed 4 years ago

a561a844

Don't handle bias inside cudnn_convolution* (#31524)

zasdfgbnm committed 4 years ago

34561dad

no_grad, enable_grad: support for decorating generator functions (#31792)

peterbell10 committed 4 years ago

5d80f634

Add missing TORCH_CUDA_API annotation to throw_nccl_error (#31157)

ezyang committed 4 years ago

58cffbff

Newer Older