strisunshinewentingwang/pytorch

Pull Requests Commits

Tensor.contiguous, Tensor.is_contiguous batch rule (#47621)

zou3519 committed 3 years ago

fc24d065

Revert to use NCCL 2.7.8-1 (#47638)

mrshenli committed 3 years ago

6c815c71

[ONNX] Bool inputs to index_put updated symbolic (#46866)

shubhambhokare1 committed 3 years ago

1abe6e5a

[ONNX] Export and shape inference for prim uninitialized in If subblock (#46094)

neginraoof committed 3 years ago

da2e2336

[TB][embedding supporting] Modify histogram to accept multipy types to skip Castop and avoid OOMing in Castop

Peiyao Zhou committed 3 years ago

4078f446

[hotfix] fix collect_env not working when torch compile/install fails (#47752)

Rong Rong committed 3 years ago

513f62b4

Added CUDA support for complex input for torch.inverse #2 (#47595)

IvanYashchuk committed 3 years ago

a1db5b0f

[FX] Fix uses not updating when erasing a node (#47720)

James Reed committed 3 years ago

dbfee42a

[FX] Add a bunch of docstrings (#47719)

James Reed committed 3 years ago

d1351c66

Revert D23632280: [c10d] switch ProcessGroup::Work to be managed by intrusive_ptr

wanchaol committed 3 years ago

dac01921

Revert D24667128: [c10d] switch Store to be managed by intrusive_ptr

wanchaol committed 3 years ago

1f946e94

Revert D24667127: [c10d] switch ProcessGroupNCCL:Options to be managed by intrusive_ptr

wanchaol committed 3 years ago

2204374f

Convert from higher order functions to classes in tools.codegen.gen (#47008)

ezyang committed 3 years ago

0c64f9f5

Fix classmethod override argument passing. (#47114)

hameerabbasi committed 3 years ago

d478605d

[quant][graphmode][fx] Support standalone_module_class (#47705)

jerryzh168 committed 3 years ago

1239d067

Allow for string literal return during symbolic tracing (#47618)

Ansley Ussery committed 3 years ago

4cb73f5a

Stop including TypeDefault.h from MPSCNNTests.mm (#46998)

ezyang committed 3 years ago

48ed577f

[fbcode][pytorch mobile] Create model reader utilities.

xcheng16 committed 3 years ago

88ec72e1

Revert D24859919: [pytorch][PR] Grammatically updated the tech docs

ejguan committed 3 years ago

5647f0ca

[c10d] switch ProcessGroupNCCL:Options to be managed by intrusive_ptr (#47075)

wanchaol committed 3 years ago

ae5c2feb

[c10d] switch Store to be managed by intrusive_ptr (#47074)

wanchaol committed 3 years ago

0cfe3451

[c10d] switch ProcessGroup::Work to be managed by intrusive_ptr (#44046)

wanchaol committed 3 years ago

0650a616

Unbreak backward compatibility tests (#47726)

ngimel committed 3 years ago

cbf439ca

[vulkan] Apply new changes to vulkan api v1 (#47721)

IvanKobzarev committed 3 years ago

bfec376e

Use local env for building CUDA extensions on Windows (#47150)

peterjc123 committed 3 years ago

d73a8db2

Fix output type of torch.max for Tensor subclasses. (#47110)

hameerabbasi committed 3 years ago

7908bf27

Fixed einsum compatibility/performance issues (#46398)

heitorschueroff committed 3 years ago

a5c65b86

[vulkan] tentative fix for conv2d_pw, and fix checks for addmm (#47723)

SS-JIA committed 3 years ago

51a661c0

Support default args in symbolic tracing (#47615)

Ansley Ussery committed 3 years ago

e914a1b9

Add max_src_column_width to autograd profiler (#46257)

Will Feng (DPER) committed 3 years ago

a5e9fa1b

Newer Older