strisunshinewentingwang/pytorch

Pull Requests Commits

[TensorExpr] Move lowerings registration from kernel.cpp to lowerings.cpp. (#65553)

Mikhail Zolotukhin committed 2 years ago

765b6a90

[TensorExpr] Move 'compute*' functions to operators/... (#65552)

Mikhail Zolotukhin committed 2 years ago

015e0079

[TensorExpr] Port NNC lowerings to the new registry mechanism. (#65551)

Mikhail Zolotukhin committed 2 years ago

3a0165da

[TensorExpr] Add a skeleton for a registry of NNC lowerings. (#65550)

Mikhail Zolotukhin committed 2 years ago

eee9ad0f

[TensorExpr] Kernel: make prim::ConstantChunk handled like other ops. (#65549)

Mikhail Zolotukhin committed 2 years ago

d84191fc

[Static Runtime] Make module_ optional in StaticModule (#65882)

Hao Lu committed 2 years ago

a6ad2b41

slow_conv2d grad_input: avoid dispatch in parallel region (#65725)

peterbell10 committed 2 years ago

08df4c2b

Make JIT Aliasing Test Less Brittle (#65493)

Jiewen Tan committed 2 years ago

6502fb89

[QPL] move metadata logging to markerEnd for model run QPL (#65451)

linbinyu committed 2 years ago

4f5ea598

[caffe2] fix LLVM-12 nullptr-with-nonzero-offset UBSAN error (#65506)

Igor Sugak committed 2 years ago

2481c064

Migrate THCCachingHostAllocator to ATen (#65746)

peterbell10 committed 2 years ago

f6dfac69

[ONNX] Enable export of __xor_ (#64042) (#64581)

BowenBao committed 2 years ago

d3979034

[ONNX] Fix inplace fill_ dtype export mismatch (#64233) (#64580)

BowenBao committed 2 years ago

e598ba2e

[ONNX] Update doc and error message for indexing export (#64290) (#64579)

BowenBao committed 2 years ago

89cbe622

[ONNX] Fix remainder export (#64230) (#64578)

BowenBao committed 2 years ago

d4ff344f

Add onnx test for batched_nms (#53175) (#64381)

BowenBao committed 2 years ago

0f0ef4fe

[ONNX] Fix gather squeeze axis in constant folding (#63588) (#64379)

BowenBao committed 2 years ago

7e15f2dd

[ONNX] Fix cuda test case (#63597) (#64378)

BowenBao committed 2 years ago

41bdfe39

[ONNX] Fix input sequence for pad op (#60554) (#64377)

BowenBao committed 2 years ago

2d61009f

Fix empty size constant creation (#63607) (#64376)

BowenBao committed 2 years ago

f17ee368

[ONNX] Update instance_norm implementation and support training (#60538) (#64375)

BowenBao committed 2 years ago

84190daf

[fx2trt][quant] Add lowering support for per channel quantization in fx2trt (#64787)

jerryzh168 committed 2 years ago

3d6d4f43

Delete rouge cu102 windows builds (#65961)

malfet committed 2 years ago

207fefc9

Clarified difference in behavior of `empty_strided` and `as_strided` (#64568)

ysiraichi committed 2 years ago

b3da2afe

Revert D31137652: [pytorch][PR] Skip failing tests when LAPACK and MAGMA are not available

suo committed 2 years ago

22f36353

Implement n-dimensional hermitian FFTs (#63890)

peterbell10 committed 2 years ago

6285348f

Add __module__ to torch.dtype.__dict__ (#65182)

nelhage committed 2 years ago

70f9f58a

[PyTorch][Edge] Fix inefficiency in objLoaderMobile (#65710)

swolchok committed 2 years ago

38c77539

[MicroBench] Added a micro benchmark for prefix sum (#65790)

navahgar committed 2 years ago

8f398325

[ci] fix softmax bc check (#65952)

suo committed 2 years ago

24f59fa2

Newer Older