pytorch/xla

Pull Requests Commits

Add aten::addcmul and aten::addcdiv operations.

dlibenzi committed 6 years ago

3d08bc1d

Added aten::clamp and aten::clamp_ operations.

dlibenzi committed 6 years ago

e3fd28f2

Added aten::neg operation.

dlibenzi committed 6 years ago

0a95cf81

Make more methods in XLATensor static

asuhan committed 6 years ago

f3d7c8be

Added aten::pow and aten::sqrt operators.

dlibenzi committed 6 years ago

6c39cf31

Added aten::exp and aten::log operators.

dlibenzi committed 6 years ago

bb483ba1

Add adaptive_avg_pool2d_backward to ATen XLA tensor

asuhan committed 6 years ago

1390a01e

Add adaptive_avg_pool2d to ATen XLA tensor

asuhan committed 6 years ago

a9e148d6

Added aten::sin and aten::cos operators.

dlibenzi committed 6 years ago

17dfe4e7

Add comments about the IR Node/Value shape APIs.

dlibenzi committed 6 years ago

df9e7376

Add native_batch_norm[_backward] to ATen XLA tensor

asuhan committed 6 years ago

3a34e168

Use the ATen batch_norm when training is not set

asuhan committed 6 years ago

88c4fd8f

Remove patches which have landed

asuhan committed 6 years ago

ff3b7c04

Re-enable BN test.

dlibenzi committed 6 years ago

9f5e0c0a

Use Value::shape() instead of implicit cast and Node::shape()

asuhan committed 6 years ago

b94039b0

Import new TF with HLO evaluator fix.

dlibenzi committed 6 years ago

0de962c4

Simplify output shape for BatchNormForward

asuhan committed 6 years ago

ec123533

Force overwrite of const& tensor in our _s_copy_from() API.

dlibenzi committed 6 years ago

ad6e234e

Rename patch to prepare for FB version coming in.

dlibenzi committed 6 years ago

c9277c13

Support installing Bazel from latest versions.

dlibenzi committed 6 years ago

8d7eb740

New changes in PT turned the Storage constructor we were using, into wanting an allocator (even though never used).

dlibenzi committed 6 years ago

baa0b921

Allow test filtering via -F option.

dlibenzi committed 6 years ago

e666bcd1

Make Node::ReturnOp(s) check the number of outputs

asuhan committed 6 years ago

eba59e27

Fix typo: GetTestTesor -> GetTestTensor

asuhan committed 6 years ago

bf00c4c6

Add batch_norm to ATen XLA tensor

asuhan committed 6 years ago

013b090f

Update to latest TF

asuhan committed 6 years ago

2c7de4c3

Move hasher implementation to CPP file to reuse the hash combiner, instead of copying its code.

dlibenzi committed 6 years ago

b062728b

Drop explicit initialization.

dlibenzi committed 6 years ago

d9189335

Added PT select() support in XLA.

dlibenzi committed 6 years ago

b7e688cc

Use dereference operator in IR Values.

dlibenzi committed 6 years ago

c193320e

Older