pytorch/xla

Pull Requests Commits

Added automatic conversion to IR Output for the IR NodeOperand.

dlibenzi committed 7 years ago

6a295e91

Use the proper output index for applying operations.

dlibenzi committed 7 years ago

1d3c4388

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

dlibenzi committed 7 years ago

62d6e804

dlibenzi committed 7 years ago

cc779e86

Added multi-ops IR Node return facility.

dlibenzi committed 7 years ago

cf68d03a

Added maxpool and log softmax backward tests.

dlibenzi committed 7 years ago

7eace363

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

dlibenzi committed 7 years ago

7d07cd4f

Added convolution backward test.

dlibenzi committed 7 years ago

b9015b1b

Correctly set sizes, strides and num elements.

dlibenzi committed 7 years ago

5119438f

Account for undefined tensors coming out of ATEN results.

dlibenzi committed 7 years ago

d20e7431

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

dlibenzi committed 7 years ago

9e2e7459

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

dlibenzi committed 7 years ago

b15504d0

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

dlibenzi committed 7 years ago

3dacb078

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

dlibenzi committed 7 years ago

ddeccf09

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

dlibenzi committed 7 years ago

d6021358

Generate debug metric code for intercepted functions.

dlibenzi committed 7 years ago

35a7583f

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

dlibenzi committed 7 years ago

9478c8a9

Address review comments

asuhan committed 7 years ago

e1d2be53

Add XLA lowerings to AtenXlaType

asuhan committed 7 years ago

d7781763

Fix XlaTensorDevice

asuhan committed 7 years ago

c795036a

Make scripts/gen.py use newly added enum values

asuhan committed 7 years ago

756e7d6c

The XLA TensorImpl needs an empty Storage, as other APIs in PT wants to access it to fetch things like device or data type.

dlibenzi committed 7 years ago

b5c2a57a

Add XLA tensor implementation method override, as this is used by make_variable() on XLA tensors, and if not overridden causes crashes trying to access not existent Storage.

dlibenzi committed 7 years ago

caa3cc79

Added a README and removed the now-applied PR patch.

dlibenzi committed 7 years ago

f3e974f9

Add infrastructure to apply outstanding PT PRs.

dlibenzi committed 7 years ago

a6a14550

Fix code generation for new IntArrayRef type.

dlibenzi committed 7 years ago

f86fb971

Update the PyTorch commit id to latest (#184)

asuhan committed 7 years ago

Verified 27bc6de8

Make setuptools compile in parallel. (#182)

dlibenzi committed 7 years ago

e883c8bf

Update the PyTorch commit id to latest (#180)

asuhan committed 7 years ago

1efd052b

Batch tensor conversions to minimize the RPC count, compile, execute,… (#179)

dlibenzi committed 7 years ago

Verified 533ec15c

Newer Older