strisunshinewentingwang/pytorch

Pull Requests Commits

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

fduwjj committed 1 year ago

92923aca

[export] Cleanup constraints (#102666)

angelayi committed 1 year ago

7a569f86

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

xndcn committed 1 year ago

bebb8b7c

Add size info to collective logs (#100413)

kwen2501 committed 1 year ago

a548fab8

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

wanchaol committed 1 year ago

c5d4ee2d

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

XiaobingSuper committed 1 year ago

49cd184f

Update XLA pin (#102446)

malfet committed 1 year ago

49d0d1d7

Allow more inserts before reIndexTopology (#102312)

zdevito committed 1 year ago

b9294c7c

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

yipjustin committed 1 year ago

6b8e68ce

Use size in python list (#102538)

pandaninjas committed 1 year ago

20ca994a

support ConvBinaryInplace in Inductor cpp wrapper (#101394)

chunyuan-w committed 1 year ago

0d2e7a18

Add ngimel to Core Reviewers (#102668)

malfet committed 1 year ago

cdfba6fc

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

ZainRizvi committed 1 year ago

c84f246c

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

PaliC committed 1 year ago

693114c0

Clean up mypy (#102555)

wconstab committed 1 year ago

398a5f4d

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

shaoyf42 committed 1 year ago

8d7e0823

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

ezyang committed 1 year ago

e03800a9

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

huydhn committed 1 year ago

cba40049

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

tugsbayasgalan committed 1 year ago

d9f75dde

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

huydhn committed 1 year ago

04c1c2b7

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

pritamdamania87 committed 1 year ago

9a2df0a5

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

zou3519 committed 1 year ago

fc31b3a1

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

anijain2305 committed 1 year ago

65631d45

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

heidongxianhua committed 1 year ago

213e10dc

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

lantiankaikai committed 1 year ago

17166c25

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

cyyever committed 1 year ago

3ae42cb7

[Replicate] Add unit test with replicate param names (#102401)

rohan-varma committed 1 year ago

0ecca122

Run slow gradcheck on the newer G5 runner (#102496)

huydhn committed 1 year ago

9331b7fa

Faketensor hpu device normalization (#102512)

VRSinghHabana committed 1 year ago

c27cefcc

Enable hipSOLVER in ROCm builds (#97370)

alugorey committed 1 year ago

eaffd988

Newer Older