strisunshinewentingwang/pytorch

Pull Requests Commits

[BE] Remove bool dtype from `masked_scatter` (#98015)

malfet committed 1 year ago

a1dc2b17

using accumulate type to do the computation of mean reduce(CPU) (#97351)

XiaobingSuper committed 1 year ago

26a90fb9

Fix format bug in NT docs (#97998)

drisspg committed 1 year ago

a5b6f10c

separate deterministic scatter_add as a helper function (#97922)

yuguo68 committed 1 year ago

fae28fcd

[Vulkan] Fix divide-by-zero with padded tensors (#97698)

manuelcandales committed 1 year ago

99f25c29

[ci][easy] Only print remaining logs if test step ran (#97713)

clee2000 committed 1 year ago

38207a9e

[dtensor] switch mean to use reduction linear (#97996)

wanchaol committed 1 year ago

1b323b31

Move functional collectives to the right namespace (#97793)

Rodrigo Kumpera committed 1 year ago

184bfbc3

Revert "[BE][autograd Function] Raise an error if input is returned as-is and saved for forward or backward in setup_context (#97212)"

pytorchmergebot committed 1 year ago

45acfc85

[dynamo] profiler.record_function on all dynamo_timed functions (#96495)

davidberard98 committed 1 year ago

c218309f

[PyTorch] Optimize TupleType::annotation_str_impl for small tuples (#97910)

swolchok committed 1 year ago

ca135ed6

Flip switch (#97993)

eellison committed 1 year ago

cadccf0d

make triton uses the wheel script now (#97995)

ezyang committed 1 year ago

f2e6b083

Skip test_batch_norm in test_jit_fuser_te for asan (#98016)

clee2000 committed 1 year ago

1f85390e

[vmap] Fix index_select support when dim is negative (#97916)

zou3519 committed 1 year ago

7bb5fb3c

Revert "Disable dynamo tracing torchrec.distributed (#97824)"

pytorchmergebot committed 1 year ago

7868e4b4

Fix module backward pre-hooks to actually update gradient (#97983)

soulitzer committed 1 year ago

ee1c539e

[dynamo 3.11] fix push null timing in resume functions (#96504)

williamwen42 committed 1 year ago

06d677f4

[dynamo 3.11] properly determine cell/freevar index in bytecode_transformation.py (#96503)

williamwen42 committed 1 year ago

5b6e4c48

[dynamo 3.11] properly copy free/cell vars in eval_frame.c (#96501)

williamwen42 committed 1 year ago

ba52268d

[inductor] fix TritonTemplateCaller.__str__ (#97578)

shunting314 committed 1 year ago

c681c52e

[dynamo 3.11] fix eval_frame.c debug prints for 3.11 (#96500)

williamwen42 committed 1 year ago

c905251f

fix functional collective to not generate getattr node (#97924)

wanchaol committed 1 year ago

848bf810

[CUDA][CUDA 11] Remove more CUDA 11 version checks (#92934)

eqy committed 1 year ago

2fddcf0f

[inductor] test codegen with dynamic shapes (#96934)

nkaretnikov committed 1 year ago

90f69cad

distinguish mutability of StorageImpl::data_ptr() member (#97651)

mikey dagitses committed 1 year ago

da28af32

set num_warps to at least 4 (#97950)

ngimel committed 1 year ago

35090b86

Fix TorchScript support in `as_nested_tensor` (#97960)

rusty1s committed 1 year ago

19706356

Compare `len(fw_derivatives)` with 0 w/o using `not` (#97953)

crcrpar committed 1 year ago

b235e1f7

Run the benchmark suite with dynamic batch only (#97912)

ezyang committed 1 year ago

97fc8ea5

Newer Older