pytorch/xla

Pull Requests Commits

Move logical_element_type to XLATensor::Data

asuhan committed 6 years ago

becc590e

Don't run shape inference in ComparisonOp

asuhan committed 6 years ago

bed4412d

Add assertions in MakeTensorFromXlaLiteral

asuhan committed 6 years ago

33e8d072

Only store logical element type instead of shape

asuhan committed 6 years ago

c27c6099

Use S32 as physical representation for comparison results

asuhan committed 6 years ago

3a423708

Share comparison op implementation with translator

asuhan committed 6 years ago

6816c5d8

Add ATen XLA comparison methods against scalar

asuhan committed 6 years ago

01ca6755

Add more ATen XLA comparison methods

asuhan committed 6 years ago

e5f0eaac

Added aten::slice operator.

dlibenzi committed 6 years ago

7276fd73

Added aten::softmax operator.

dlibenzi committed 6 years ago

21832040

Added aten::min and aten::max operations.

dlibenzi committed 6 years ago

d1888521

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

dlibenzi committed 6 years ago

51cad05f

Add dim method test

asuhan committed 6 years ago

cc545a67

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

Newer Older