strisunshinewentingwang/pytorch

Pull Requests Commits

[FSDP] Include duplicate parameters and modules when calling named_parameters and named_modules (#98912)

fegin committed 1 year ago

8e328762

trivial data reads to const_data_ptr (#99004)

mikey dagitses committed 1 year ago

a44813e6

Adds 3D attn_mask support to merge_masks() for Multihead Attention fast path (#98991)

LucasLLC committed 1 year ago

35c6547f

make ATen/native/cuda/WeightNorm.cu data_ptr-correct (#99006)

mikey dagitses committed 1 year ago

bae304a5

Enable fused optimizer for DP (#98270)

rohan-varma committed 1 year ago

bba20908

Enable test_matmul_cuda UTs for ROCm (#98797)

BLOrange-AMD committed 1 year ago

079452ea

Move/Fix FakeTensor logic for detecting multiple fake modes (#97186)

eellison committed 1 year ago

fc53472c

[unwind] clarify warnings (#99005)

zdevito committed 1 year ago

82b8764b

[FSDP] Full precision in eval mode (#97645)

rohan-varma committed 1 year ago

d8b09b01

_mm_prefetch is for Intel, changed to __prefetch for Arm64 (#96638)

Blackhex committed 1 year ago

c7431061

trivially migrate std::transform output to mutable_data_ptr (#98756)

mikey dagitses committed 1 year ago

7a77961d

[Dynamo] Fix Lazy Module initialization with constant arg (#98996)

yanboliang committed 1 year ago

e20981bd

[functorch] typo in merge rule's github handle (#99052)

kshitij12345 committed 1 year ago

7692243e

Revert "[core][pruning][be] Rename sparsifier folder to pruner (#98758)"

pytorchmergebot committed 1 year ago

dda7ce4b

[inductor] Support IndexPutFallback in cpp_wrapper (#98972)

desertfire committed 1 year ago

e5501a96

AOTAutograd: fix 'Trying to backward through the graph a second time' error (#98960)

bdhirsh committed 1 year ago

670c5cf9

trivially convert std::copy output to mutable_data_ptr (#98755)

mikey dagitses committed 1 year ago

39438c68

Enable passing a dict of module names: log level to set_logs python api (#98989)

mlazos committed 1 year ago

3a400a5a

[quant][pt2e][improvement] Remove the need to annotate all nodes with default annotation (#99001)

jerryzh168 committed 1 year ago

6a568779

Add test for squeeze.dims shape function (#98144)

vivekkhandelwal1 committed 1 year ago

f501234b

[fix] update the condition for aliveness of TensorWrapper (#98748)

kshitij12345 committed 1 year ago

2c337dd9

[pt2] recursive IR check (#98887)

xw285cornell committed 1 year ago

0ff3059a

convert trivial assignments to use mutable_data_ptr (#98752)

mikey dagitses committed 1 year ago

1854e8ac

Use the same python version in MacOS workflows and add more debug messages (#98902)

huydhn committed 1 year ago

388d2692

[ROCm] Sync updates in hipify_torch to Pytorch hipify utils for ROCm. (#93169)

wenchenvincent committed 1 year ago

8155b72c

Disable logging in pattern matcher calls to AotAutograd (#98936)

jansel committed 1 year ago

8a6dd0dc

inductor: relax the dynamic variable check for cpu dynamic test case (#98815)

XiaobingSuper committed 1 year ago

8a3f1be8

Support module hooks in UnspecializedNNModuleVar (#98540)

wconstab committed 1 year ago

a408ed24

Revert "[quant][fix] Compare resnet with quantizer api with the prepare_fx and decomposed convert flow (#98905)"

pytorchmergebot committed 1 year ago

731590ba

Make update_expected not fail on one missing file (#98982)

wconstab committed 1 year ago

296822c4

Newer Older