strisunshinewentingwang/pytorch

Pull Requests Commits

[FSDP] Test `use_orig_params=True` in `test_fsdp_ignored_modules.py` (#90290)

awgu committed 1 year ago

3b578edd

Fix uniform ref implementation (#90094)

yanboliang committed 1 year ago

25f39c1b

ShapeEnv.create_symbolic_sizes_strides_storage_offset (#89962)

ezyang committed 1 year ago

a1ab06ab

reland #89222: [Composable API] replicate: change to per module call, remove mark_root_module() (#90254)

Charlie Yan committed 1 year ago

e818c366

[FSDP] Fix accidental change in `_test_fsdp_parity` (#90252)

awgu committed 1 year ago

bd9ad89a

Log1p for complex in CPU (#89691)

mfkasim1 committed 1 year ago

ce212628

[dtensor] handle the case where output of op is Optional[Tensor] (#90241)

wanchaol committed 1 year ago

9e314bd8

Use Sized not Iterable to test for len (#90182)

ezyang committed 1 year ago

eace0848

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

Newer Older