strisunshinewentingwang/pytorch

Pull Requests Commits

[FSDP] Fix bug in determining whether parameters need to be materialized

jamesr66a committed 2 years ago

723f8b9a

Revert "remove dead torch_pb.h library (#97323)"

pytorchmergebot committed 2 years ago

12da0c70

Revert "move caffe2/proto/ to its own Bazel package (#97324)"

pytorchmergebot committed 2 years ago

b531eb97

Revert "cleanup caffe2 cc_proto_library (#97325)"

pytorchmergebot committed 2 years ago

91a3040b

Implement last dim split_with_sizes for NT (forward only, non-SymInt-ified) (#97446)

jbschlosser committed 2 years ago

0d66db1b

[ci] disable some dtensor tests (#97358)

wanchaol committed 2 years ago

37f7c13b

Revert "remove dead proto_convert library (#97322)"

pytorchmergebot committed 2 years ago

13fbf932

Rewrite NCCL watchdog to more reliably throw timeout (#97066)

kwen2501 committed 2 years ago

95e8d0c3

[Vulkan] Fix static analysis errors in vulkan_quantized_api_test.cpp (#97400)

manuelcandales committed 2 years ago

416bac5b

[DTensor] default value for DTensor ops on non-participating devices (#95852)

XilunWu committed 2 years ago

c2d75082

[DTensor] implement aten.equal sharding prop (#97170)

XilunWu committed 2 years ago

103f4c99

Rename torch._inductor.triton_ops.autotune to torch._inductor.triton_heuristics (#95558)

jansel committed 2 years ago

5f57b363

update flatten.py docstring (#97276)

neonwatty committed 2 years ago

f0649d47

update internal triton (#97422)

yuguo68 committed 2 years ago

a3b30c50

Remove non existent files in multigpu tests (#97393)

Catherine Lee committed 2 years ago

29c061bb

Fix potential naming clash when writing traces with tensorboard_trace_handler (#97392)

huydhn committed 2 years ago

4a88f71f

[inductor] Fix a multi-gpu context error (#97398)

desertfire committed 2 years ago

d499b7d7

vmap support for linalg.lu_factor (#94328)

isdanni committed 2 years ago

7711d247

build C++ extensions on windows with /std:c++17 (#97413)

mikey dagitses committed 2 years ago

bdaf4025

[inductor] handle integer `Symbol`s in `is_integer_type` (#97217)

nkaretnikov committed 2 years ago

feace5d6

[inductor] fix cpp legalize bf16 reduction (#97228)

Valentine233 committed 2 years ago

a331cd43

[FSDP][optim_state_dict] Consolidate the arguments and logic of optim_state_dict and optim_state_dict_to_load (#96534)

fegin committed 2 years ago

580b4702

Retry download and install NDK when testing Android (#97067)

huydhn committed 2 years ago

1fb1c6e1

DCE inference graphs too (#97275)

ezyang committed 2 years ago

37faa488

Only warn once for TypedStorage deprecation (#97379)

kurtamohler committed 2 years ago

fbc803df

Fix Device Idx Setting (#97399)

eellison committed 2 years ago

b507d7d7

Sort: Use cub::WarpMergeSort for small sorts (32 < n <= 128) (#96223)

peterbell10 committed 2 years ago

5d8c7e7e

[PyTorch] Add annotation_str benchmark (#96496)

swolchok committed 2 years ago

3b545920

[vision hash update] update the pinned vision hash (#97396)

pytorchmergebot committed 2 years ago

a34d35d5

Fix typo under torch/csrc/jit/passes directory (#97222)

kiszk committed 2 years ago

62ecfa8b

Older