pytorch/xla

Pull Requests Commits

Add thnn_conv2d_backward to ATen XLA tensor

asuhan committed 6 years ago

25399bc1

Add avg_pool2d_backward to ATen XLA tensor

asuhan committed 6 years ago

2b4bfc3a

Fix formatting in third_party/xla_client/util.h

asuhan committed 6 years ago

49946e08

Do max pool with dilation through ATen interop

asuhan committed 6 years ago

83db8fe3

Hash IR Node metadata.

dlibenzi committed 6 years ago

f4f4bad5

dlibenzi committed 6 years ago

36ed6c17

Add fast has generation from IR Node objects to be able to quickly detect graph changes.

dlibenzi committed 6 years ago

947dfafa

Create device shape out of IR Node shape.

dlibenzi committed 6 years ago

3c7dcf94

Add dereference operator to access IR Node pointer from NodeOperand, and added more testing for the IR Node replace operation.

dlibenzi committed 6 years ago

d1edc278

Added automatic conversion to IR Output for the IR NodeOperand.

dlibenzi committed 6 years ago

6a295e91

Use the proper output index for applying operations.

dlibenzi committed 6 years ago

1d3c4388

Turn IR operation to act on NodeOperand in order to support multi-output IR Node within XLA tensors.

dlibenzi committed 6 years ago

62d6e804

dlibenzi committed 6 years ago

cc779e86

Added multi-ops IR Node return facility.

dlibenzi committed 6 years ago

cf68d03a

Added maxpool and log softmax backward tests.

dlibenzi committed 6 years ago

7eace363

Wrap the generated regsitration function with an std::once_flag.

dlibenzi committed 6 years ago

7d07cd4f

Added convolution backward test.

dlibenzi committed 6 years ago

b9015b1b

Correctly set sizes, strides and num elements.

dlibenzi committed 6 years ago

5119438f

Account for undefined tensors coming out of ATEN results.

dlibenzi committed 6 years ago

d20e7431

Drop the USE_XRT env config, as there is no more XLA option.

dlibenzi committed 6 years ago

9e2e7459

Use at::Scalar instead of double for IR Scalar Node.

dlibenzi committed 6 years ago

b15504d0

Make C++ test utility more resilient to different inputs (XLA vs ATEN tensors, variables or raw tensors).

dlibenzi committed 6 years ago

3dacb078

Move the variable creation of ToTensor() APIs out to the Python binding as otherwise autograd barks.

dlibenzi committed 6 years ago

ddeccf09

Fix PT/XLA tensor shape generation to create tuple when input is tuple.

dlibenzi committed 6 years ago

d6021358

Generate debug metric code for intercepted functions.

dlibenzi committed 6 years ago

35a7583f

Remove dummy file as the folder has a README file now.

dlibenzi committed 6 years ago

9478c8a9

Address review comments

asuhan committed 6 years ago

e1d2be53

Add XLA lowerings to AtenXlaType

asuhan committed 6 years ago

d7781763

Fix XlaTensorDevice

asuhan committed 6 years ago

c795036a

Make scripts/gen.py use newly added enum values

asuhan committed 6 years ago

756e7d6c

Newer Older