strisunshinewentingwang/pytorch

Pull Requests Commits

[GHA][BE] Remove unneeded shellcheck suppressions (#74308)

malfet committed 2 years ago

Verified ef8995ff

First step of refactor lower passes (#74219)

Shirong Wu committed 2 years ago

5de5dd2d

[Profiler] Switch to thread local subqueues to reduce lock contention. (#74151)

Taylor Robie committed 2 years ago

daf959c8

Remove sync in embedding (#70943)

zasdfgbnm committed 2 years ago

11dc1581

remove _lazy_init() in rebuild full params (#74263)

zhaojuanmao committed 2 years ago

52c10952

make sharding strategy configurable and support zero2 algorithm (#73819)

zhaojuanmao committed 2 years ago

5e39d949

Revert D34886988: [Quant][core][gpu][improvement] Refactored implementation for conv2d_cudnn to use packed parameters (Reland PR#73510)

David Dang committed 2 years ago

3fe16060

Revert D34886987: [Quant][core][refactorization] Refactored qconv_unpack.cpp into an implementation file and higher level call registration and definition file (Reland #73773)

David Dang committed 2 years ago

5853c8a9

[DDP][Tests] Fix weight sharing test (#74252)

rohan-varma committed 2 years ago

bbf4a998

[DDP] Generalize activation checkpoint tests (#74130)

rohan-varma committed 2 years ago

a5ea3b7f

[PyTorch] Add Tensor.is_nested (#73999)

swolchok committed 2 years ago

d4a44300

Revert "Update tls logic to work better with guarded call (#73925)"

zou3519 committed 2 years ago

a9d9f91f

[ATen][AMD] revert a change of USE_DIRECT_NVRTC (#74194)

shintaro-iwasaki committed 2 years ago

102add66

Preserve codegen on fx graph in transformer (#74189)

842974287 committed 2 years ago

f98b316f

Dispatch from torch.Tensor.to_sparse_coo to to_sparse

cpuhrsch committed 2 years ago

4de9cb9a

[PyTorch] Add unit test for c10::Synchronized<T> (#74110)

dhruvbird committed 2 years ago

606c26d3

[PyTorch] Use c10::Synchronized<T> in RWSafeLeftRightWrapper (#74109)

dhruvbird committed 2 years ago

977cc480

[PyTorch] Update Synchronized<T>::withLock() to return the type/value from the aceepted callable (#74108)

dhruvbird committed 2 years ago

e1eb876a

[PyTorch] [Model Tracer] Use c10::Synchronized<T> for build features tracer (#74107)

dhruvbird committed 2 years ago

44ef979d

[PyTorch] [Model Tracer] Use c10::Synchronized<T> for custom classes tracer (#74106)

dhruvbird committed 2 years ago

3674f92d

[PyTorch] [Model Tracer] Use c10::Synchronized<T> for kernel dtype tracer (#74105)

dhruvbird committed 2 years ago

cf461a8b

Use assertEqual consistently in test_sparse_csr.py

cpuhrsch committed 2 years ago

a346a181

Excluding ASAN and periodic jobs from slow job calculation (#74253)

atalman committed 2 years ago

2e15e16f

Support running pipelines on main in .jenkins and tools

musebc committed 2 years ago

0effac2b

[Quant][core][refactorization] Refactored qconv_unpack.cpp into an implementation file and higher level call registration and definition file (Reland #73773) (#74227)

dzdang committed 2 years ago

ce7041d2

[Quant] qconv: fix xnnpack operator caching (#74217)

digantdesai committed 2 years ago

f8c1acea

Add operator selection ability to gen_unboxing (#74271)

priyaramani committed 2 years ago

92751498

Improve error message of loading saved TS module out of support window (#74228)

gmagogsfm committed 2 years ago

d5744f47

[Quant][core][gpu][improvement] Refactored implementation for conv2d_cudnn to use packed parameters (Reland PR#73510) (#74220)

dzdang committed 2 years ago

a2f701d4

[PyTorch GPU Allocator] Better use of blocks with rounding of allocation sizes (#74213)

banitag1 committed 2 years ago

ac3effd1

Newer Older