strisunshinewentingwang/pytorch

Pull Requests Commits

[LTC] Tweak LazyGraphExecutor for XLA

alanwaketan committed 2 years ago

4c87c88a

Fix XLA dynamo CI (#90229)

JackCaoG committed 2 years ago

2ea32f41

Add sharding strategy to fully_shard (#90192)

mrshenli committed 2 years ago

5d6aa99c

Add a repro for fully_shard _unshard error (#90190)

mrshenli committed 2 years ago

e4670885

[Composable API] `replicate`: add support for DDP args (#89243)

yhcharles committed 2 years ago

0f274ed3

[FSDP][optim_state_dict][1/N] Restructure _optim_state_dict to prepare the support of use_orig_param (#89898)

fegin committed 2 years ago

72fdfad4

Simplify by using yield from (#90160)

kit1980 committed 2 years ago

2b20a3d3

Fix issue 38095 TODOs in NCCL tests (#90033)

kit1980 committed 2 years ago

54858cce

[NNC] Use New PassManager for LLVM >= 15 (#89978)

davidberard98 committed 2 years ago

7571134f

Assume that co_firstlineno is always defined (#90180)

ezyang committed 2 years ago

5de5c5e4

workaround for indexing formulas with negative terms (#89933)

Natalia Gimelshein committed 2 years ago

1ea20cdb

fix c10::detail::integer_iterator for C++17 (#90174)

mikey dagitses committed 2 years ago

368a1cbd

Light refactor to how we get shape_env for graph lowering (#90139)

voznesenskym committed 2 years ago

5423c2f0

Fix missing line in XLA backend after mergebot + ghstack gap (#90197)

voznesenskym committed 2 years ago

32639a82

[LTC] Restore default ctor for LazyTensor (#90086)

alanwaketan committed 2 years ago

7e034193

[Composable API] `replicate`: change to per module call, remove `mark_root_module()` (#89222)

yhcharles committed 2 years ago

65a0dcff

Revert "as_strided: Fix default storage_offset for reference implementation (#89513)"

pytorchmergebot committed 2 years ago

8845a8f8

[ONNX] Fix concat with empty tensors (#87620)

Thiago Crepaldi committed 2 years ago

6d794f6a

Remove deprecated usage of is_pod/is_pod_v (#88918)

Lukas N Wirz committed 2 years ago

301d9c05

[4/4][DataPipe] Remove iterator depletion in Zipper (#89974)

ejguan committed 2 years ago

b1eb42bc

as_strided: Fix default storage_offset for reference implementation (#89513)

peterbell10 committed 2 years ago

eded97ac

Remove deprecated flatten_params_wrapper.py from lintrunner config (#90154)

mrshenli committed 2 years ago

199b8b60

Fix fully_shard error when policy is not provided (#90151)

mrshenli committed 2 years ago

7a08261a

Added vectorized flip for uint8 (#90013)

vfdev-5 committed 2 years ago

777ac632

[Inductor] handle non-positive exponents in `Pow` (#90146)

nkaretnikov committed 2 years ago

226e803e

Use dynamo fake tensor mode in aot_autograd, move aot_autograd compilation to lowering time [Merger of 89672 and 89773] (#90039)

voznesenskym committed 2 years ago

41c3b41b

Revert "Use dynamo fake tensor mode in aot_autograd, move aot_autograd compilation to lowering time [Merger of 89672 and 89773] (#90039)"

pytorchmergebot committed 2 years ago

4648baa9

[codemod][llvm15] LLVM-15 fixes for caffe2/test/cpp/jit/test_module_api.cpp (#89938)

r-barnes committed 2 years ago

a580a634

Fix warning: use of bitwise '&' with boolean operands (#90131)

zasdfgbnm committed 2 years ago

d6c8603b

[Doc][Distributed] Add missing functions to distributed.rst (#89905)

yaox12 committed 2 years ago

57bb4cd0

Older