strisunshinewentingwang/pytorch

Pull Requests Commits

[Dynamo, Compiled] Save some python overhead when calling compiled function with many tangents (#118730)

wschin committed 203 days ago

11f2e8be

[FSDP] Added warning about unsupported double backwards (#120926)

awgu committed 203 days ago

0b18ed1c

Don't aggressively rewrite asserts for symbolic expressions (#120564)

tugsbayasgalan committed 203 days ago

f01a23d0

[dynamo] Reorder logs (#116106)

angelayi committed 203 days ago

c844b377

Exclude operators that produce unbacked symbols (#120917)

ezyang committed 203 days ago

9fc56f82

Replace TTIR string parsing with structured MLIR walk in Triton kernel mutation analysis (#120476)

aakhundov committed 203 days ago

ea7149aa

Fix guard for SUPPORTED_NODES (#120798)

aorenste committed 203 days ago

8861507b

Add sparse compressed meta tensor support (#120707)

pearu committed 203 days ago

b8e6ca6f

Make SparseCsr a functionality dispatch key (#120703)

pearu committed 203 days ago

70d4d109

expose nested header to wheel (#120603)

leizhenyuan committed 203 days ago

eee040c9

Support higher order op functionalization in predispatch IR (#115314)

tugsbayasgalan committed 204 days ago

c646030c

Move VariableInfo into its own file to avoid circular dependency (#120732)

xmfan committed 204 days ago

82b35619

[PT2D] Make the speedup benchmark works with DDP + CompiledAutograd (#120454)

fegin committed 204 days ago

8c2e5699

Add verbose parameter to torch.hub.list (#120717)

cyyever committed 204 days ago

77ef9d40

Revert "[dynamo] Reorder logs (#116106)"

pytorchmergebot committed 204 days ago

63b25949

[Inductor][cuDNN] Disable tf32 in `test_mutate_view_for_conv_output` (#120953)

eqy committed 204 days ago

86e6497c

Update xfails for scaled_dot_product_attention (#120928)

davidberard98 committed 204 days ago

6ed26392

Add _assert_scalar and teach Inductor to codegen it (#114148)

ezyang committed 204 days ago

2a08a517

Add test to check that COW inputs are not materialized (#119507)

kurtamohler committed 204 days ago

77aea289

Avoid COW materialization in `at::parallel_for/parallel_reduce` (#120455)

kurtamohler committed 204 days ago

13a54ce2

delete useless cast_outputs call in unary_op_impl_float_out (#120486)

Shan19900305 committed 204 days ago

d053dcfa

[dynamo] Reorder logs (#116106)

angelayi committed 204 days ago

c5472628

Enable TORCH_TRACE by default in all Tupperware like environments (#120915)

ezyang committed 204 days ago

02a410ee

support bool as Scalar Type in TorchScript (#113835)

majian4work committed 204 days ago

518a23bb

[executorch hash update] update the pinned executorch hash (#120747)

pytorchupdatebot committed 204 days ago

2e84d01d

Revert "[Dynamo] Fix inspect.getattr_static doesn't work well for torch.utils._cxx_pytree.PyTreeSpec (#120812)"

pytorchmergebot committed 204 days ago

65d56868

[onnxrt, dynamo] Enable custom ONNX model transforms in `onnxrt` dynamo backend (#120854)

wschin committed 204 days ago

e49f31ca

fix the scale dot attention doc (#120859)

lancerts committed 204 days ago

67c97a9a

Ban reset_to_zero argument to triton.autotune in user defined kernels (#120938)

oulgen committed 204 days ago

b35551f3

Change FakeTensor serialization to consider only an _active_ FakeTensor mode (#120848)

masnesral committed 204 days ago

06f8af30

Newer Older