strisunshinewentingwang/pytorch

Pull Requests Commits

inductor: eliminate meaningless copy (#102089)

blzheng committed 2 years ago

e59db086

[Quant][PT2E][Inductor] Lower quantized conv to Inductor (#101164)

Xia-Weiwen committed 2 years ago

ce9923a1

add foreach support for custom device (#102047)

heidongxianhua committed 2 years ago

b088ff46

[Dynamo] Correct UserDefinedObjectVariable.var_getattr on function/method type (#102580)

yanboliang committed 2 years ago

9fa82c90

[TP] Use Stride inferred from local tensor in to_local bwd (#102630)

fduwjj committed 2 years ago

92923aca

[export] Cleanup constraints (#102666)

angelayi committed 2 years ago

7a569f86

[inductor] use native fetch_add function for trivial types (#101931)

xndcn committed 2 years ago

bebb8b7c

Add size info to collective logs (#100413)

kwen2501 committed 2 years ago

a548fab8

[dtensor][simple] fix some comments (#102661)

wanchaol committed 2 years ago

c5d4ee2d

inductor: improve the index range check for index_expr vec check (#102263)

XiaobingSuper committed 2 years ago

49cd184f

Update XLA pin (#102446)

malfet committed 2 years ago

49d0d1d7

Allow more inserts before reIndexTopology (#102312)

zdevito committed 2 years ago

b9294c7c

[pytorch-vulkan] aten::uniform (#102431)

yipjustin committed 2 years ago

6b8e68ce

Use size in python list (#102538)

pandaninjas committed 2 years ago

20ca994a

support ConvBinaryInplace in Inductor cpp wrapper (#101394)

chunyuan-w committed 2 years ago

0d2e7a18

Add ngimel to Core Reviewers (#102668)

malfet committed 2 years ago

cdfba6fc

Improve time savings calculation math for test reordering (#102411)

ZainRizvi committed 2 years ago

c84f246c

Adds script to generate alerts for failing jobs (#102002)

PaliC committed 2 years ago

693114c0

Clean up mypy (#102555)

wconstab committed 2 years ago

398a5f4d

[c10d] Add is_backend_available for c10d backend. (#101945)

shaoyf42 committed 2 years ago

8d7e0823

Add torch._utils.render_call, improve printoptions (#102623)

ezyang committed 2 years ago

e03800a9

Run libtorch in 2 shards (manual sharding) (#102554)

huydhn committed 2 years ago

cba40049

[export] Add aot_export 1/N (#101490)

tugsbayasgalan committed 2 years ago

d9f75dde

Try to build the Docker image if it doesn't exist (#102562)

huydhn committed 2 years ago

04c1c2b7

[RFC] Add method to DDP to check for backward finalization. (#100773)

pritamdamania87 committed 2 years ago

9a2df0a5

Allow existing "Python RAII guards" to be used as context managers (#102579)

zou3519 committed 2 years ago

fc31b3a1

[benchmarks] Use train mode for accuracy checks for HF models (#102578)

anijain2305 committed 2 years ago

65631d45

fix bug in trace model when out-operator has more than one output (#101563)

heidongxianhua committed 2 years ago

213e10dc

python_arg_parser to allow fake tensor element in symint_list when in dynamo mode #95424 (#97508)

lantiankaikai committed 2 years ago

17166c25

adjust header inclusions in C10 as sugguested by IWYU (#102467)

cyyever committed 2 years ago

3ae42cb7

Newer Older