strisunshinewentingwang/pytorch

Pull Requests Commits

Add _int_mm to expose cuBLAS int8@int8 -> int32 matmul (#94339)

cpuhrsch committed 1 year ago

1fe2a9d1

make overriding operator warning message only print once (#95179)

heidongxianhua committed 1 year ago

32558910

[NCCL] (re-open) Optionally avoid `recordStream` calls in `ProcessGroupNCCL` (#89880)

eqy committed 1 year ago

29f9a702

[meta_tensor] polish error strings in meta registrations (#95052)

1SAA committed 1 year ago

f43ce955

[CI] Do not compare two eager run results against fp64 result (#95616)

desertfire committed 1 year ago

fa5a4b0d

dynamo export should be able to export identity function (#94962)

qihqi committed 1 year ago

34617d7e

Re-enable a FX-to-ONNX kwargs Test (#94763)

wschin committed 1 year ago

868640e0

Update `fx.pass.graph_drawer` usage doc to draw fx graph (#95534)

wenzhe-nrv committed 1 year ago

8dfac7b8

Cleanup Windows warning suppression in CMake and fix some warnings in the source code (#94927)

cyyever committed 1 year ago

f27e09de

Revert "[Functional Collectives] Migrate DeviceMesh::all_reduce to use functional all_reduce. (#95009)"

pytorchmergebot committed 1 year ago

d950f455

Add bfloat16 support to upsample (#95500)

jamaliki committed 1 year ago

1cf11c1c

Fix split_module bug (#95493)

RenfeiChen-FB committed 1 year ago

c44a7330

[Quant] Fix setting fixed qparams for inner LSTM ops (#95537)

andrewor14 committed 1 year ago

a3b505c5

Fix typos in documents under torch (#95597)

kiszk committed 1 year ago

31ce32b0

[dynamo] Fix keyword argument name of all_dim (#95600)

kiszk committed 1 year ago

3beb6445

Fix potential deadlock when recording memory traces (#95273)

zdevito committed 1 year ago

4f84c57c

Fix typos under torch/_inductor directory (#95601)

kiszk committed 1 year ago

9a4cb9bc

Expose more headers for extensions. (#95447)

donnyyou committed 1 year ago

5d70ee93

suppress nvfuser loading warning when we disable nvfuser (#95603)

cyyever committed 1 year ago

c1fa403e

Fix double-a typo (#95470)

rasbt committed 1 year ago

97ec340f

[MPS] Add roll op (#95168)

qqaatw committed 1 year ago

4930ae7f

Revert "Disable MacOS M1 test jobs (#95509)"

pytorchmergebot committed 1 year ago

448c97ca

Implement sparse semantics support in gradcheck (2nd try) (#95405)

pearu committed 1 year ago

b89fda51

[inductor] Allow list of decompositions to be overridden (#95468)

tbennun committed 1 year ago

ea367347

add/add_ for compressed sparse inputs: bypass BLAS in some trivial cases (#95293)

nikitaved committed 1 year ago

325b4366

Deepcopy output node metadata (#95426)

angelayi committed 1 year ago

d301caa8

Avoid copies in matmul (#76828)

lezcano committed 1 year ago

b3175ae9

[inductor] enable `test_grid_sampler_2d_dynamic_shapes` (#95575)

nkaretnikov committed 1 year ago

d83a14e7

Don't create large intermediary tensors in the backward of matmul (#95261)

lezcano committed 1 year ago

03cc0f58

[MPS][BE] Introduce xfail (#95045)

malfet committed 1 year ago

fd8367a7

Newer Older