strisunshinewentingwang/pytorch

Pull Requests Commits

Support Custom Memory Plan in CudaCachingAllocator (#83837)

o-hanna committed 2 years ago

Verified e3f54d46

track allocate/free event sequences in CudaCachingAllocator (#81753)

o-hanna committed 2 years ago

Verified f50273f2

[pocket fft] turning on pocketfft flag (#81670)

mcr229 committed 2 years ago

0b3a239e

Follow GitHub relabeling of oncall: fx for test owners (#81821)

ezyang committed 2 years ago

5b88a207

[CUDA graphs] Clear autocast amp cache (#81558)

Aidyn-A committed 2 years ago

e9d07bd4

Add native impl for group norm on quantized CPU for channels-last inputs (#70520)

Xia-Weiwen committed 2 years ago

c1795977

[fix] build failure in python 3.10 (#81812)

kshitij12345 committed 2 years ago

94090150

corrects typo in quantization docs (#81687)

zaouk committed 2 years ago

2fb2740e

Revert "Revert "Refactored prim utils into _prims_utils folder (#81746)

Chillee committed 2 years ago

a5fb41e3

Revert "[Reland] Add should_traverse_fn to torch.fx.node.map_aggregate (#81695)"

pytorchmergebot committed 2 years ago

75aa049a

Add cpu/gloo tests for sharded tensor distributed checkpoint (#80997)

zzzwen committed 2 years ago

b8f9751f

[GHA] Re-enable backwards compatibility test! (#81795)

janeyx99 committed 2 years ago

1bf5ce47

[JIT] Modify is_mutable in FunctionSchema and SchemaInfo to have SchemaArgument parameter instead of index (#81784)

goldenxuett committed 2 years ago

c9497886

[JIT] Modify SchemaCheckMode to utilize SchemaInfo and fix training op tests accordingly (#81783)

goldenxuett committed 2 years ago

41d054ab

[JIT] Enchance training ops check to be more inclusive and account for possible pybind exceptions (#81782)

goldenxuett committed 2 years ago

21a4be34

Enable reentrant dispatch for decompositions (#81598)

davidberard98 committed 2 years ago

08b9544e

[composite compliance] test_operator correctness (#81600)

kshitij12345 committed 2 years ago

8b5685da

Upgrade mypy in lintrunner to 0.960 (#81775)

ezyang committed 2 years ago

d681ed56

Revert "FIX make sure we import the correct object from multiprocessing (#53282)"

pytorchmergebot committed 2 years ago

3c9479dc

correct argument name in docs (#81485)

kylematoba committed 2 years ago

66cf1b64

Re-enable C++ doc generation (#81719)

jbschlosser committed 2 years ago

8573da59

Add partitioned nvFuser executor with ATen fallbacks (#81043)

IvanYashchuk committed 2 years ago

a3d5d2dd

Expose Lint for AliasDb (#81579)

wschin committed 2 years ago

767facdb

[fix] allow saving python attr on Tensor and Parameter via torch.save (#81616)

kshitij12345 committed 2 years ago

f3f8d96e

[Profiler] Add pattern to detect if TF32 is available but not used (#81273)

davidchencsl committed 2 years ago

26ecd67d

Add pattern to detect for loop indexing into tensor (#81056)

davidchencsl committed 2 years ago

6e621d02

[composite compliance] check output of forward-ad with subclass args against regular tensor (#81464)

kshitij12345 committed 2 years ago

706b420a

Add test that all BackendComponents are covered by toString (#81713)

ezyang committed 2 years ago

b24c1022

Fix printing of DispatchKey in operator not found message (#81637)

ezyang committed 2 years ago

41d8c24a

FIX make sure we import the correct object from multiprocessing (#53282)

tomMoral committed 2 years ago

e103d6af

Older