strisunshinewentingwang/pytorch

Pull Requests Commits

[Profiler] Encapsulate callback handle management. (#83892)

Taylor Robie committed 2 years ago

c26b53f6

Removing multigpu 10.2 . Using 11.6 cuda for multigpu tests instead (#84286)

atalman committed 2 years ago

ddd841b3

Revert "Test distributed backends in parallel (#84034)"

pytorchmergebot committed 2 years ago

772721a4

modify split_by_tags to retain output order (#84136)

ihoffman1028 committed 2 years ago

20018aa7

Add nvfuser support for squeeze (#84117)

IvanYashchuk committed 2 years ago

90161c23

[Nested Tensor]Clean up offsets (#84145)

drisspg committed 2 years ago

174c3c68

Test distributed backends in parallel (#84034)

huydhn committed 2 years ago

3ae5be74

[ONNX] refactor test_pytorch_onnx_onnxruntime_cuda.py (#84218)

titaiwangms committed 2 years ago

641c3952

[Easy][FSDP] Update `StateDictType` doc (#84200)

awgu committed 2 years ago

b8ee8101

[Easy][FSDP] ufmt `_optim_utils.py` (#84199)

awgu committed 2 years ago

7f58db74

[ONNX] Add script/trace different flatten and move optional type tests to runtime (#83184)

titaiwangms committed 2 years ago

5bceaadb

Fix the edge case when y = 0 in kl_div (#82714)

lezcano committed 2 years ago

b106a04d

Fix issue in softmax.cu with transformer error when mask seqlen > 1024 (#83639)

erichan1 committed 2 years ago

b182f081

Fix split torch_function handling (#83866)

peterbell10 committed 2 years ago

897907d4

[c10d] Introduce dist.get_local_rank, dist.get_global_rank and dist.get_global_ranks (#82134)

kumpera committed 2 years ago

65dc5dd3

Set default value for nccl make MAX_JOBS if ProcessorCount returns 0 (#84231)

mrshenli committed 2 years ago

56a37ea1

[Easy][FSDP] Fix sharded optim state dict doc formatting (#84198)

awgu committed 2 years ago

f0efc1c2

Update README.md (#84263)

Pompey21 committed 2 years ago

546d6822

Revert "Re-land sym_numel (#82374) (#82726) (#82731) (#82855)" (#84207)

Krovatkin committed 2 years ago

44a97533

Revert "Use self-hosted runner for viable/strict update (#84249)"

pytorchmergebot committed 2 years ago

60f47cb0

Use self-hosted runner for viable/strict update (#84249)

janeyx99 committed 2 years ago

acd6ca8c

[PT] Allowing deepcopy in unitialized parameter (#83809)

shenxiul committed 2 years ago

ec714e33

Vectorize conversions to BFloat16 on CPU (#80906)

peterbell10 committed 2 years ago

856a7d94

only run the circleci mac/ios jobs on prs (#84227)

clee2000 committed 2 years ago

7a14c56b

[Nested Tensor] fix from_padded bug (#84217)

drisspg committed 2 years ago

71369051

[fx] Make get_isolated_graphmodule accept tracing mode. (#84238)

zhxchen17 committed 2 years ago

df98c529

[functorch] fix multinomial (#83838)

samdow committed 2 years ago

399b1eb8

[reland] Make allreduce compatible with make_fx (#84221)

mrshenli committed 2 years ago

34e5b099

[fx] Make wrapped_fn also work for non-mutating passes. (#84232)

zhxchen17 committed 2 years ago

a402e100

Fix typo (#83802)

hhy3 committed 2 years ago

8aba2535

Newer Older