pytorch/xla

Pull Requests Commits

Added aten::argmin and aten::argmax operations.

dlibenzi committed 6 years ago

f401497c

Added at::expm1 and at::log1p operators

vishwakftw committed 6 years ago

9bbc08dd

WIP: Added aten::mean operation.

dlibenzi committed 6 years ago

01f0de84

Pin PT commit ID in order to be able to work until a clear fix is in place.

dlibenzi committed 6 years ago

9d933687

Added acos, cosh, asin, asinh unary ops

vishwakftw committed 6 years ago

43c03ed1

Add permute to ATen XLA tensor

asuhan committed 6 years ago

1c631871

Added aten::where operation.

dlibenzi committed 6 years ago

f49bbb09

Add unsqueeze to ATen XLA tensor

asuhan committed 6 years ago

d3d7bf40

Added aten::gather operation.

dlibenzi committed 6 years ago

42b405ce

Consolidate some code for boring 1:1 XLA mapping ops.

dlibenzi committed 6 years ago

1c1a1c20

Don't depend on translator header in data_ops header

asuhan committed 6 years ago

7563cd1c

Allow graph parameters for reshapes target size

asuhan committed 6 years ago

39dc105b

Factor out XLA lowering in squeeze operator

asuhan committed 6 years ago

9e8c7538

Use streaming in GetCanonicalDimensionIndex error

asuhan committed 6 years ago

8ffc6d4f

Add squeeze to ATen XLA tensor

asuhan committed 6 years ago

0ea2be6c

Pin build to confirmed working PyTorch commit id

asuhan committed 6 years ago

7eb178b3

Do not die of consuminsm.

dlibenzi committed 6 years ago

31a5d89c

Check XLATensorImpl is always contiguous

asuhan committed 6 years ago

112d7e32

Fix breakage due to c10 Storage API change.

dlibenzi committed 6 years ago

c35e2bde

Add aten::abs operator

vishwakftw committed 6 years ago

03560382

Migrate to using the full path for include files.

dlibenzi committed 6 years ago

97547eee

Make BuildComparisonOp more compact

asuhan committed 6 years ago

b07909a3

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

Older