strisunshinewentingwang/pytorch

Pull Requests Commits

[MPS] Fix `test_embedding_dense_backward` (#88847)

malfet committed 2 years ago

62ef15e3

[Dynamo] Add complete support for Tensor.is_contiguous (#88407)

yanboliang committed 2 years ago

b30222e0

Fix cupti search path in CMake (#88657)

dzhulgakov committed 2 years ago

ae01615d

Symbolic shape: sym_floor , sym_sqrt, sym_int (#88760)

SherlockNoMad committed 2 years ago

d9ad08ce

[Inductor] Fix lowmem_dropout() missing 1 required positional argument: 'p' (#88716)

yanboliang committed 2 years ago

cc04cf50

[ONNX] Create common ExportTestCase base class (#88145)

BowenBao committed 2 years ago

500fd655

[ONNX] Improve diagnostic message formatting (#87830)

BowenBao committed 2 years ago

20ae19aa

[ao] qconfig_mapping_utils.py fixing public v private (#87517)

HDCharles committed 2 years ago

a6610faa

Have kernel names include fused ops (#88624)

mlazos committed 2 years ago

c1553880

[ao] fuser_method_mappings.py fixing public v private (#87516)

HDCharles committed 2 years ago

ad2eba80

[xnnpack][lite-int][on-device] rebuild serialized modules at runtime (#88780)

mcr229 committed 2 years ago

37b468ac

Use run_test in MPS (#88829)

clee2000 committed 2 years ago

de38c876

[inductor] Remove import check for fast_flush (#88812)

bertmaher committed 2 years ago

1ae772a6

[xnnpack][on-device] compiler --> executor object (#88779)

mcr229 committed 2 years ago

3a4e8736

sub setup.py install -> develop (#88507)

msaroufim committed 2 years ago

394b998d

[xnnpack][on-device] executor class (#88778)

mcr229 committed 2 years ago

d5e1e2f0

Revert "[Inductor] Build FX Linear + Permute Vertical Fusion in Inductor (#88566)"

pytorchmergebot committed 2 years ago

29550e2c

[DataPipe] Deprecating drop_empty_batches from Filter and other functional APIs (#88693)

ejguan committed 2 years ago

90cf14dd

Bring Unfold/Fold param doc order in line with code (#88819)

felixdivo committed 2 years ago

98ecd065

[14/N] Refactor _new_process_group_helper() to remove repeated code (#88351)

H-Huang committed 2 years ago

1d54ce9d

Add warnings & regressions info text (#88837)

williamwen42 committed 2 years ago

4bcf2c53

[LTC] Make ComputePostOrder accept const T pointers (#88773)

alanwaketan committed 2 years ago

3b8245ab

[Inductor] Build FX Linear + Permute Vertical Fusion in Inductor (#88566)

jiawenliu64 committed 2 years ago

48b58930

Revert "Symintify `broadcast_to` (#88776)"

pytorchmergebot committed 2 years ago

d157fca5

[FSDP][Perf] Do not call `pad` in no-padding case (#88769)

awgu committed 2 years ago

6bf2776a

[dynamo] `VariableTracker.call_method` requires a name (#88311)

bertmaher committed 2 years ago

d3178465

[nnc] Disable opaque pointers mode in LLVM backend to allow getPointerElementType (#88798)

bertmaher committed 2 years ago

1e4079a4

Change TORCH_INTERNAL_ASSERT to TORCH_CHECK and add a nice error message (#88804)

PanosAntoniadis committed 2 years ago

656d0de6

Switch to setup-nvidia action (#88757)

huydhn committed 2 years ago

79b049af

Make TorchElastic timer importable on Windows (#88522)

malfet committed 2 years ago

f98edfcc

Newer Older