strisunshinewentingwang/pytorch

Pull Requests Commits

sparse_mask: remove redundant mask.coalesce() in to_dense_backward (#92001)

nikitaved committed 1 year ago

91920ee6

Fixup torch jit with some initializers and moves (#92037)

Skylion007 committed 1 year ago

b9182cbb

generate set_device call to ensure context existence (#92055)

ngimel committed 1 year ago

5625f521

[Inductor] Support vectorized transpose in CPP backend (#91532)

jgong5 committed 1 year ago

7c641eaa

[inductor] Reduce device context manager overhead (#91045)

peterbell10 committed 1 year ago

eece6da1

Revert "[Modes] Add assert that the mode isn't already on the stack (#90770)"

pytorchmergebot committed 1 year ago

db466ae0

[FSDP] Do not clean FQNs even for `use_orig_params=True` (#91767)

awgu committed 1 year ago

a383789f

[FSDP] Test `use_orig_params=True`, `no_sync()`, mixed precision (#91193)

awgu committed 1 year ago

7f50ff16

[FSDP] Re-support model dtype change after FSDP init (#91192)

awgu committed 1 year ago

e5503ace

[BC-Breaking] Separate `stream_id`, `device_index`, and `device_type` in `pack` and `unpack` for `Streams` (#81596)

eqy committed 1 year ago

e096d2db

[dynamo] delegate handling of len() of TensorVariable to size(0) (#92016)

ydwu4 committed 1 year ago

a2368a7c

optimize sampled_addmm performance on CPU (SparseCSR) (#90978)

mingfeima committed 1 year ago

3ab58fd5

Cleanup some unused includes (#91961)

peterbell10 committed 1 year ago

81f7c406

Use c10 math constants consistently in Math.h (#91967)

peterbell10 committed 1 year ago

8acf0e62

Revert "Add check-sparse-tensor-invariants flag to Context. (#90849)"

pytorchmergebot committed 1 year ago

c7a22bb7

[functorch] Fix proxy unwrapping for cond(). (#91907)

zhxchen17 committed 1 year ago

05d0c4ce

Fix `_foreach_norm` on some tensor sizes (#91844)

milesial committed 1 year ago

a76bc410

properly convert fill value to x dtype in constant_pad (#92045)

ngimel committed 1 year ago

44413f25

[cuBLAS][TF32] Fix TF32 get/set test when `TORCH_ALLOW_TF32_CUBLAS_OVERRIDE` is set (#92052)

eqy committed 1 year ago

fb38b9ff

Fix for swap_custom_module_to_observer doing duplicate swaps on the same node.target (#91905)

harshitkhaitan committed 1 year ago

ffbd13b6

[testing] add ErrorInputs for `adaptive_{avg, max}_poolnd` (#90924)

khushi-411 committed 1 year ago

ccd8b66b

Handle tensor default func args when inlining (#90575)

wconstab committed 1 year ago

6cfaa922

Propagate sources in VariableBuilder and add SuperSource (#91729)

wconstab committed 1 year ago

8e2e648f

Add `device_idx` to `free_fn` in `CUDAPluggableAllocator` (#91398)

Emilio Castillo committed 1 year ago

07e595e8

[vision hash update] update the pinned vision hash (#91744)

pytorchmergebot committed 1 year ago

723d7641

sparse_mask: faster, with support for uncoalesced mask (#91964)

nikitaved committed 1 year ago

18677d52

[FSDP] Clarify `MixedPrecision` docs (#91974)

awgu committed 1 year ago

33052659

Implement hybrid sparse to/from dense conversions. (#90177)

alexsamardzic committed 1 year ago

8612ec5b

[Quant] make x86 the default quantization backend (qengine) (#91235)

Xia-Weiwen committed 1 year ago

e1bcbbf1

[functorch] Fix map() operator behavior. (#91906)

zhxchen17 committed 1 year ago

5766764d

Newer Older