strisunshinewentingwang/pytorch

Pull Requests Commits

add shape check for random_samples in fractional_max_pool{2d|3d} (#89992)

mingfeima committed 1 year ago

c6942dbb

replace memset with value-initialization (#90048)

mikey dagitses committed 1 year ago

be5108d5

[Quant] Add fused linear-leaky_relu op for onednn backend (#88478)

Xia-Weiwen committed 1 year ago

97e47a52

[ONNX] Add src/index dynamic axes support for aten::scatter_add (#90090)

titaiwangms committed 1 year ago

41bfa49d

Revert "[PT-D][Composability][1/N] Upstream NamedOptimizer from TorchRec (KeyedOptimizer in TR) (#89480)"

pytorchmergebot committed 1 year ago

176b962f

Add factory functions to python frontend (#89230)

rdspring1 committed 1 year ago

3c9431f5

Revert "as_strided: Fix default storage_offset for reference implementation (#89513)"

pytorchmergebot committed 1 year ago

e645771e

Improve Autograd Documentation Clarity (#89401)

ArekSredzki committed 1 year ago

44dac51c

[Vulkan] Enable QInt8 and QInt32 quantization (#89788)

manuelcandales committed 1 year ago

49ccc41d

[FSDP()] Fix `fully_shard` fwd hook registration (#90201)

awgu committed 1 year ago

45b40be0

fix: Moving operators to FuncTorchBatchedDecomposition (#89762)

srossross committed 1 year ago

2b7fcfa3

fix: update error when tensor escapes vmap (#89077)

srossross committed 1 year ago

bb673fb1

[dtensor] remove torchgen function schema and parse manually (#90106)

wanchaol committed 1 year ago

2c2cce73

remove backward hook in memory_tracker (#90143)

zhaojuanmao committed 1 year ago

a0c7b888

Fix issue 38095 TODO in onnx/test_utility_funs.py (#90085)

kit1980 committed 1 year ago

6bbcd025

[ReduceOp] ameliorate custom `__eq__` (#90088)

crcrpar committed 1 year ago

50891612

[dynamo] Rewrite addcdiv in dynamo to its constituent ops (#90227)

mlazos committed 1 year ago

2d9267ba

Fix exception causes in fx, nn and onnx packages (#90134)

cool-RR committed 1 year ago

77f9b2e8

[PT-D][Composability][1/N] Upstream NamedOptimizer from TorchRec (KeyedOptimizer in TR) (#89480)

fduwjj committed 1 year ago

31ec1a1e

[ao][ns] PNP demo for exposing arbitrary model transforms (#90153)

HDCharles committed 1 year ago

cee396fa

Disallow registering meta function for CompositeImplicitAutograd ops (#90222)

SherlockNoMad committed 1 year ago

42705bd7

pad low precision matmuls when requested (#90235)

ngimel committed 1 year ago

a88400e0

as_strided: Fix default storage_offset for reference implementation (#89513)

peterbell10 committed 1 year ago

ba70a8be

Functionalization: skip meta block computation if compute_reference_meta is false (#90219)

isdanni committed 1 year ago

05ccbd6d

Assert there are no outstanding side effects before calling cond (#90208)

ezyang committed 1 year ago

962ebe88

Revert "[Composable API] `replicate`: change to per module call, remove `mark_root_module()` (#89222)"

pytorchmergebot committed 1 year ago

0d8e53df

[vision hash update] update the pinned vision hash (#90239)

pytorchmergebot committed 1 year ago

73565ce3

Revert "[Composable API] `replicate`: add support for DDP args (#89243)"

pytorchmergebot committed 1 year ago

3749b9dc

TorchDynamo: always convert flexiblelayout to be FixedLayout when given a stride_order (#89904)

XiaobingSuper committed 1 year ago

2597d5d7

[inductor] Add test_ops_gradients running with inductor (#89792)

desertfire committed 1 year ago

29233a18

Older