strisunshinewentingwang/pytorch

Pull Requests Commits

Fix DDP incompatibility issue with nn.MultiheadAttention.

Guanheng Zhang committed 4 years ago

cefc4d5c

Remove _dequantize_per_tensor (#26681)

jerryzh168 committed 4 years ago

3f72bcfc

Make `is_optional` check more robust (#26312)

davidriazati committed 4 years ago

d0fff0eb

Add doc building instructions

David Riazati committed 4 years ago

5cc35348

Un-hardcode epsilon constant in FoldConvBatchNorm2d.

Mikhail Zolotukhin committed 4 years ago

eddda3af

Register values listed in __constants__ as attributes of the Module. (#26581)

Mikhail Zolotukhin committed 4 years ago

6c758ff2

Remove _dequantize_per_channel in the pattern (#26680)

jerryzh168 committed 4 years ago

52b69fbc

Trivial quantized torch.mean implementation

James Reed committed 4 years ago

cf272d43

Enable hub tests on MacOS (#26697)

ailzhang committed 4 years ago

9f1da984

Setting automatic default selection for ONNX IR v4 semantics in ONNX export API (#26146)

Spandan Tiwari committed 4 years ago

af3b15b7

Add traces to specialize_autograd and lower_grad_of (2nd try)

Krovatkin committed 4 years ago

8b126022

Expands TestAutogradDeviceType (#26708)

Mike Ruberry committed 4 years ago

a172fbf9

Remove duplicate calculation of output shape (#26684)

dskhudia committed 4 years ago

fa7b621a

Use noop observer to pass dtype for dynamic quantization (#26709)

Dmytro Dzhulgakov committed 4 years ago

128a65e2

Speed up an integer to the power of a positive integer on CPU (#26020)

xuhdev committed 4 years ago

ae0732cd

allow building docker without torchvision (#26168)

David Pollack committed 4 years ago

66d27504

Add tests for C++ functional cosine_similarity and pairwise_distance, and clean up functional test code (#26559)

Will Feng committed 4 years ago

3cae3021

Updating submodules

svcscm committed 4 years ago

714b05e4

Don't generate named tensor functions to RegistrationFunctions.h (#26685)

zou3519 committed 4 years ago

ff78d743

Typo fix (#26417)

ezyang committed 4 years ago

05f70818

resolve ignored module method type annotations (#26683)

Elias Ellison committed 4 years ago

efaa65dd

remove tools/setup_helpers/cudnn.py (#25876)

xuhdev committed 4 years ago

5e5cbcee

Add warning to anomaly_mode doc fix #26408 (#26615)

albanD committed 4 years ago

9f3351de

Vectorize unary operator erfinv (#26629)

pearu committed 4 years ago

cf1dbc79

Add derivative for cholesky_inverse (#26451)

vishwakftw committed 4 years ago

c6432909

Move the CUDA implementation of trunc to ATen. (#25423)

xuhdev committed 4 years ago

7bdc0c13

Automatic update of fbcode/onnx to 23bb6ea1a71f08e200114a153f48bd7adb66d486 (#26441)

houseroad committed 4 years ago

d6ee5849

C++ API parity: at::Tensor::set_data

pbelevich committed 4 years ago

450504cd

Use optimized graph in Inline (essentially, making Inline recursive now). (#26489)

Mikhail Zolotukhin committed 4 years ago

2cf1183e

Add 'optimized_graph' to Function. (#26488)

Mikhail Zolotukhin committed 4 years ago

c522b635

Older